Zone 2 Training: What the Evidence Shows

The thing nobody warns you about is the humiliation of the small hill. You have read the posts, you have bought into the premise, you have set the watch to buzz at you when you cross the line, and you go out on a Tuesday evening intending to do the easy thing — the famously easy thing, the thing that is supposed to be so gentle it barely counts — and within nine minutes you are standing at the bottom of an incline you have driven over a thousand times without noticing, doing arithmetic. If you keep the cadence you will go over the number. If you drop to a walk you will be, in some sense you cannot articulate but feel keenly, cheating. You compromise: a sort of stiff-hipped shuffle, arms held oddly, breath deliberately slowed as though the watch can hear you. At the top the strap says the number spiked anyway, and stayed there ninety seconds, and you spend the rest of the session auditing your own cardiovascular system, wondering whether the spike ruined it, whether "it" was ever a discrete thing that could be ruined, and whether the strap is even right.

It probably is not, exactly. That is the first honest thing anyone should tell you. The second is that the line the watch is defending was drawn by a convention, not by your body, and that different conventions draw it in different places. The third is more interesting: a great deal of the research everyone cites when they talk about zone 2 was not conducted on people like you, was not measuring the outcome you care about, and in several of its most-cited trials did not find what the summary of it implies.

Here is where the evidence actually stands. Low-intensity aerobic training is a real and useful stimulus, and the largest, longest-running descriptive body of work on it comes from endurance athletes who train very large volumes and spend most of that time going easy. When that low-intensity-heavy distribution has been tested head-to-head against other distributions in randomised trials, the results have been small, inconsistent, and dependent on how trained the athlete already was — a 2024 meta-analysis of 17 studies found a modest advantage of polarized training for peak oxygen uptake and no advantage at all for time-trial performance, time to exhaustion, or threshold speed (Sports Med 2024). When low or moderate continuous training has been compared against interval training for raising VO₂max, the interval side has generally been slightly ahead, not behind. And the popular claim that zone 2 is uniquely or optimally mitochondrial has been examined directly and rejected by a 2025 narrative review in Sports Medicine that concluded current evidence does not support it (Sports Med 2025). None of that means easy training is worthless. It means the specific story attached to it is doing more work than the data can carry.

What "zone 2" actually refers to

So: what is zone 2 training? In current popular usage it means sustained aerobic exercise performed below the lactate threshold — hard enough to be clearly exercise, easy enough that blood lactate stays near resting levels and does not climb over the course of the session. That is the definition the 2025 Sports Medicine review adopts when it examines the claim, and it is a reasonable summary of what the term has come to mean online (PMID 40560504).

The complication is that "zone 2" is a coordinate in a map, and there is more than one map. In the three-zone system used throughout the training-intensity-distribution literature, the zones are defined by two physiological landmarks: the first ventilatory threshold and the respiratory compensation threshold. Zone 1 is everything below the first threshold. Zone 2 is the band between the two thresholds. Zone 3 is everything above the upper one. In the recreational-runner trial published in International Journal of Sports Physiology and Performance, the polarized group's distribution was reported as roughly 77/3/20 across those three zones, and the comparison group's as 46/35/19 — the comparison group being the one that spent more time in zone 2, which was the whole point of calling it "between-thresholds" training (PMID 23752040). A later trial in the same journal classified intensity the same way, prescribing its polarized group 85%/5%/10% in zones 1, 2 and 3 (PMID 29952662).

Read that again and the trap becomes visible. In the literature people cite to argue for lots of easy training, zone 2 is the intensity the recommended distribution minimises — three to five percent of training time. The thing being maximised is zone 1. Advice to do more zone 2, backed by a citation to polarized training, is citing a body of work in which "zone 2" names a different intensity than the one recommended. Five-zone systems shift the boundaries again, and the heart-rate-percentage zones on a consumer watch are a fourth convention on top, anchored to an estimated maximum rather than a measured threshold.

This is not pedantry. It is why two people can both be doing "zone 2" — one cruising well under their first threshold, the other grinding along just under their second — and both be following the instructions correctly. The physiological ground truth is a pair of thresholds; the zone number is a naming scheme laid over them, and the schemes disagree.

How do you know if you're in zone 2?

How do you know if you're in zone 2? With less precision than the interface implies. There are four broad approaches, each carrying its own quantified error.

Blood lactate is the reference method, and it is where the trouble starts, because there is no single agreed procedure for turning a lactate curve into a threshold. A study of eight elite cross-country skiers compared six published methods for predicting the anaerobic lactate threshold from the same incremental test data. Agreement was defined in advance as 95% limits of agreement falling within ±0.5 mM. None was found among any of the methods, and one differed significantly from every other (Int J Exerc Sci 2019). A comparison of four methods in 17 professional soccer players reached the same verdict: they should not be used interchangeably, and the heart rate identified at threshold came out systematically higher with some than others (J Strength Cond Res 2016). A swimming study offers the partial exception worth knowing: in ten competitive swimmers, two different incremental protocols produced closely matching results — a heart-rate bias between protocols of −2 beats per minute, with a standard deviation of 8 — and most calculation methods agreed with directly measured maximal lactate steady state, though the modified maximum-distance method overestimated it (Int J Sports Physiol Perform 2024). Its title states the finding well: the method, but not the protocol, affects where the threshold lands.

Ventilatory thresholds are the other laboratory route, identified from gas exchange during a maximal incremental test, and they are what the polarized-training trials used to set their zones and control the intervention (PMID 23752040, PMID 29952662).

The talk test is the cheap one, and it is better than it sounds. A validation study manipulated the ventilatory threshold in healthy young adults — lowering it by blood donation, raising it by training — and found that the exercise intensity at the last stage where speech remained comfortable tracked the shifted threshold in both directions. It also found the failure mode: errors were biased toward passing the talk test when intensity already exceeded the threshold, and speech took longer than the two minutes used in earlier protocols to become genuinely difficult (J Cardiopulm Rehabil Prev 2008). A review in Current Opinion in Cardiology summarised it: comfortable speech is likely possible below the ventilatory or lactate threshold and not likely above it, the relationship holds across walking, jogging, cycling, elliptical and stair stepper, and the test is impractical during intervals (Curr Opin Cardiol 2014).

Which brings us to the watch, and to 220 minus age. The formula is not a measurement; it is a population regression line, and every individual sits some distance off it, so a band computed from an estimated maximum inherits that error before the strap adds its own. No study in the evidence base assembled for this piece validated the equation directly, so the honest statement is directional rather than numerical: the error in an age-estimated maximum is large enough that a zone drawn from it can sit meaningfully above or below your actual threshold, and the size of that displacement is unknown for you specifically. Hold that next to the finding that even in a laboratory, with real blood samples, six analysis methods applied to one dataset failed to agree (PMID 30761194). The number on the wrist is an estimate derived from an estimate.

Does zone 2 training work?

Does zone 2 training work? The answer depends entirely on what you mean by "work," and on who you are, and the second condition is the one most often skipped.

Almost all the direct experimental evidence on training-intensity distribution comes from trained endurance athletes. The individual-participant network meta-analysis published in Sports Medicine in 2025 pooled 13 studies and 348 athletes — 296 men and 52 women, 150 recreational and 198 competitive, mean ages 17.6 to 41.5 years, VO₂max across studies 46.6 to 68.3 mL·kg⁻¹·min⁻¹ (PMID 39888556). The earlier meta-analysis of polarized versus threshold training required participants to be endurance-trained with more than two years of experience and a VO₂max above 50 (J Strength Cond Res 2019). If you are a 52-year-old who runs three times a week, you are outside the entry criteria of the literature being quoted at you.

What high-volume low-intensity training changes in that population is real. In a 15-day live-high train-low camp, 12 elite cross-country skiers accumulated 45.9 ± 6.4 hours of low-intensity work below the first ventilatory threshold — 80% of their training — against 23.9 ± 2.8 hours and 55% for eight Nordic-combined athletes, who instead did far more strength and jump work. Roller-ski performance improved by 2.9 ± 1.6% in the skiers and worsened by 4.1 ± 2.6% in the Nordic-combined group. VO₂max rose about 3.7% in both, but oxygen uptake at the second ventilatory threshold improved only in the skiers, by 7.3 ± 5.8%, and every Nordic-combined athlete showed signs of overreaching afterwards (Front Physiol 2018). In that comparison the higher-low-intensity group improved and the other did not — though the groups were different sports doing very different amounts of strength and jump work, so the intensity distribution is only one of the things that differed. The submaximal threshold that moved is also not the same variable as maximal oxygen uptake.

Outside athletes, the interventions stop being "zone 2" in any recognisable sense. In 50 sedentary people aged 61 ± 6 with prediabetes and metabolic syndrome, dietary advice alone over 16 weeks reduced skeletal muscle citrate synthase activity by 18% ± 43% and 3-hydroxyacyl-CoA dehydrogenase activity by 23% ± 19%, while adding 32 ± 2 sessions of small-sided recreational soccer left no detectable group-level decline — though citrate synthase activity still fell in the men in both arms (Scand J Med Sci Sports 2024). Small-sided soccer is intermittent and intensity-variable; it is not a steady sub-threshold ride. In older adults, a meta-analysis of randomised trials found interval training improved peak oxygen uptake more than moderate continuous training by 1.74 mL·kg⁻¹·min⁻¹ (95% CI 0.80 to 2.69) (Exp Gerontol 2021). And in stable coronary artery disease, a six-centre trial of 382 patients found twice-weekly low-volume intervals raised VO₂peak by 2.37 mL·kg⁻¹·min⁻¹ against 1.32 for moderate steady-state exercise, an adjusted difference of 1.04 (95% CI 0.38 to 1.69, p = 0.002), with one serious adverse event possibly related (Eur J Prev Cardiol 2023).

Easy aerobic training works, in the sense that aerobic training works. What has not been demonstrated is that the sub-threshold intensity does something the other intensities cannot.

What the polarized-training literature actually found

Polarized training — a lot of zone 1, a little zone 3, almost nothing in between — is the model that gave the popular zone-2 advice its scientific cover. It has been tested repeatedly. The results are more modest and more conditional than the reputation.

The 2019 meta-analysis of randomised controlled trials found only four eligible studies and could pool just three for time-trial performance. The pooled effect favoured polarized over threshold training, effect size −0.66 (95% CI −1.17 to −0.15). Note what that rests on: two of the four studies scored 4 out of 10 on the PEDro scale and two scored 5 (PMID 29863593).

The 2024 systematic review with meta-analysis was larger — 17 studies, 437 participants — and more informative because it reported the outcomes where the advantage disappeared. Polarized training was superior for VO₂peak, standardised mean difference 0.24 (95% CI 0.01 to 0.48, p = 0.040, high certainty of evidence), but that superiority was confined to interventions shorter than 12 weeks (SMD 0.40, 95% CI 0.08 to 0.71) and to highly trained athletes (SMD 0.46, 95% CI 0.10 to 0.82). For time-trial performance the effect was essentially zero: SMD −0.01 (95% CI −0.28 to 0.25, p = 0.92); time to exhaustion, SMD 0.30 (p = 0.24); speed or power at the second threshold, SMD 0.04 (p = 0.75). The authors conclude that polarized training more effectively improves aerobic power but is similar to other distributions for aerobic capacity (PMID 38717713).

The individual-participant network meta-analysis found no difference at all between polarized and pyramidal distributions using the time-in-heart-rate-zone approach — VO₂max SMD −0.06, p = 0.68; time-trial SMD −0.05, p = 0.34 — and none between polarized and any other distribution tested. Its one interesting signal was a subgroup interaction: the VO₂max response differed between recreational and competitive athletes (SMD −0.63, p < 0.05), with competitive athletes possibly favouring polarized and recreational athletes possibly favouring pyramidal — that is, favouring more time in the between-thresholds band, not less (PMID 39888556). A 2026 Bayesian network meta-analysis was more deflationary still: no distribution showed a definite advantage over polarized for either outcome, because every 95% credible interval crossed zero. Ranking probabilities, which are not demonstrated superiority, put threshold training most likely optimal for VO₂max (rank-1 probability 65.4%, SUCRA 84.8%) and interval training most likely optimal for time trial (53.9%, SUCRA 81.5%) (J Strength Cond Res 2026).

The recreational-runner trials deserve particular attention, because recreational runners are closer to most readers than elite skiers are. Thirty endurance runners were randomised to a polarized distribution (about 77/3/20) or a between-thresholds distribution (about 46/35/19) for ten weeks, with 10 km race times measured before and after on the same course. Both groups improved significantly. The polarized group improved 5.0% and the between-thresholds group 3.6% — a difference of roughly 41 seconds — and that difference was not statistically significant. A post-hoc analysis restricted to the runners who most faithfully executed their assigned distribution did show an advantage for polarized training, 1.29 standardised effect-size units (90% CI 0.31 to 2.27, p = 0.038), but a subset selected on compliance is a weaker piece of evidence than the randomised comparison it sits inside (PMID 23752040).

The second recreational-runner trial ran polarized endurance training against a mixed high-intensity program for 12 weeks in 21 participants. The polarized group actually ran 283 minutes and 47.3 km per week, in a 74%/11%/15% distribution; the comparison group — which ran three days a week and did CrossFit on three others — did 117 minutes and 19.3 km per week in a 46%/15%/39% distribution. Five-kilometre time improved 6.21% in the polarized group and 5.49% in the comparison group, and the between-group difference was not significant (p = 0.79). Body fat fell by a similar amount in both (p = 0.88). Only VO₂max separated them, rising 4.3 mL·kg⁻¹·min⁻¹ in the polarized group versus 1.78 in the comparison group (p = 0.04, d = 0.85) (PMID 29952662). Read plainly, that trial says roughly 2.4 times the weekly running time bought an equivalent race improvement and a better laboratory number.

Zone 2 vs HIIT for VO₂max

On the question people actually want settled — zone 2 vs HIIT — the meta-analytic answer is unglamorous and has been stable for a decade: both work, both work a lot, and interval training is slightly ahead.

The 2015 Sports Medicine meta-analysis pooled 28 controlled trials and 723 healthy adults aged 18 to 45, mean age 25.1 ± 5 years with baseline VO₂max of 40.8 ± 7.9 mL·kg⁻¹·min⁻¹. Against non-exercising controls, continuous endurance training produced a possibly large benefit of 4.9 mL·kg⁻¹·min⁻¹ (95% confidence limits ±1.4) and interval training a likely large benefit of 5.5 (±1.2). Head to head, the advantage for interval training was 1.2 mL·kg⁻¹·min⁻¹ (±0.9), which the authors classify as a possibly small beneficial effect (Sports Med 2015). That is the honest headline number. Roughly one millilitre per kilogram per minute, in favour of intervals, on top of a common effect four to five times larger.

Two things sharpen it. A meta-analysis of 37 interval-training studies covering 40 training groups and 334 subjects under 45, training six to 13 weeks, found a mean VO₂max increase of 0.51 L·min⁻¹ (95% CI 0.43 to 0.60), rising to roughly 0.8–0.9 L·min⁻¹ in a subset of nine studies using longer intervals (PLoS One 2013). And a Journal of Physiology review concluded that VO₂max is generally increased more by interval training than by moderate continuous training for a given training volume, while sprint interval training and continuous training improve it similarly despite very different volumes (J Physiol 2017). Volume and intensity trade against each other; that trade is the finding, not a hierarchy of zones.

The null results matter here too. In women specifically, a systematic review and meta-analysis found no statistical difference between moderate-to-vigorous continuous training and interval training for VO₂max — mean difference −0.42 (95% CI −1.43 to 0.60) — with both improving substantially from baseline (3.20, 95% CI 2.73 to 3.67; and 3.16, 95% CI 2.09 to 4.24). Women who completed more sessions improved more, in both formats (Int J Sports Med 2023). In clinical populations the interval advantage reappears: in hypertensive patients, continuous training raised VO₂max by 1.30 mL/kg/min (95% CI 0.92 to 1.68) against controls and interval training by 4.90 (95% CI 3.77 to 6.04), while for blood pressure the two were closer — a between-group systolic difference of 1.13 mmHg (95% CI −0.01 to 2.27, p = 0.05) and a diastolic difference of 1.63 mmHg (Curr Hypertens Rep 2020). Both lowered blood pressure relative to no exercise; the between-group differences were small, and favoured intervals for diastolic.

The mitochondria claim, examined

This is the load-bearing claim in the popular case for zone 2, and it is the one that has been most directly tested and found wanting.

The 2025 Sports Medicine narrative review took the claim on explicitly. Its account of where the idea came from is the important part: the recommendation derives largely from observational data on elite endurance athletes, who perform enormous volumes of low-intensity work and who also possess high mitochondrial and fatty-acid oxidative capacity. Those two facts about elite athletes coexist; the inference that one produces the other is an inference, not a finding. The review's conclusion is that current evidence does not support zone 2 as the optimal intensity for improving mitochondrial or fatty-acid oxidative capacity, and that prioritising intensities above zone 2 appears important for cardiometabolic benefit, particularly when training volume is low (PMID 40560504).

What did the human muscle studies actually measure? Mostly PGC-1α expression, a signalling marker upstream of mitochondrial biogenesis rather than mitochondrial content itself. A 2025 meta-analysis of randomised trials found PGC-1α to be the most consistently reported marker and pooled a large effect of endurance exercise on its expression, Hedge's g = 1.17 (95% CI 0.14 to 2.19), with substantial heterogeneity (I² = 84.5%). Crucially for the zone question, the subgroup comparison found large effects for both interval and continuous endurance training — g = 1.29 and 1.01 respectively — with no significant difference between the modalities (Biomol Concepts 2025). Intensity was not the discriminating variable.

The Journal of Physiology review is compatible but shaded differently: cellular stress and the resulting signals for mitochondrial biogenesis depend largely on exercise intensity, with limited work suggesting that gains in mitochondrial content are superior after interval training when work-matched comparisons are made within the same person, and sprint interval training raising content comparably to continuous training despite much less total exercise (PMID 27748956). If there is an intensity effect on mitochondria in humans, the limited evidence points up, not down.

The prediabetes trial is the closest thing in this evidence set to a mitochondrial outcome in the population that reads longevity content: sedentary 55- to 70-year-olds, muscle biopsies before and after, enzyme activities measured directly rather than inferred. Exercise preserved citrate synthase and 3-hydroxyacyl-CoA dehydrogenase activity that energy restriction alone degraded, and the effect differed by sex — the decline in citrate synthase activity was seen exclusively in men (PMID 39604207). The intervention was recreational soccer. Nobody was watching a heart-rate strap.

Fat oxidation and the "fat-burning zone"

The fat-burning zone is real as a measurement and misleading as an instruction. There is a moderate intensity at which the absolute rate of fat oxidation peaks, genuinely higher than at very high intensities where carbohydrate dominates. The question is whether training there produces more total fat oxidation than training harder.

A randomised crossover study in 12 physically active young men tested that. Each performed continuous exercise at the intensity eliciting his highest fat oxidation rate, and the same session interrupted either by one two-minute bout at 130% of the intensity associated with VO₂max or by four 20-second bouts at 170%. Average oxygen uptake and carbohydrate oxidation did not differ. Post-exercise fat oxidation was 37.5% higher after the two-minute-interruption protocol and 50% higher after the 20-second protocol than after the uninterrupted fat-max session (Appl Physiol Nutr Metab 2022). Adding hard efforts to a moderate session increased fat oxidation afterwards, which is the opposite of what the zone framing predicts.

The narrative review's judgement on fatty-acid oxidative capacity as a training adaptation — as distinct from what you burn during one session — is the same as its judgement on mitochondria: the evidence does not establish zone 2 as optimal (PMID 40560504). And none of this bears on body composition, which is governed by energy balance over weeks and not by substrate mix over an hour. In the 12-week recreational-runner trial, body fat fell by 2.45% in the high-volume polarized group and 2.62% in the low-volume high-intensity group, with no meaningful difference between them (PMID 29952662).

How long should a zone 2 session be, and how much per week

How long should a zone 2 session be? There is no trial that answers this by randomising session duration, so the defensible answer is descriptive: here is what the studies that produced the results above actually asked people to do.

The clearest single number comes from the 12-week recreational-runner trial: the polarized group was prescribed 295 ± 67 minutes per week across five running days and completed 283 ± 75.9 minutes and 47.3 ± 11.6 km, with 74% of that time below the first ventilatory threshold (PMID 29952662). Roughly 3.5 hours a week of easy running, in an already-running population, over 12 weeks, to produce a 6.2% five-kilometre improvement that a group doing 117 minutes a week nearly matched. At the elite end, the cross-country skiers accumulated 45.9 hours of low-intensity work in 15 days (PMID 30072913) — not a lifestyle option for anyone reading this.

For the general population the useful anchors come from a different literature. The dose-response meta-analysis of non-occupational physical activity across 94 cohorts found that the steepest part of the risk gradient lies between zero and 8.75 marginal MET-hours per week — the equivalent of the recommended 150 minutes per week of moderate-to-vigorous activity — with smaller marginal differences up to 17.5 MET-hours and little beyond (Br J Sports Med 2023). In the women's meta-analysis, women who completed more sessions improved more, in both formats (PMID 37084758). And the older-adult meta-analysis found its favourable subgroups — for interval training, not for continuous work — at two sessions per week and around 40 minutes per session (PMID 33836261).

Where the longevity claim comes from — and what kind of evidence it is

The longevity case for zone 2 is built out of two literatures, neither of which studied zone 2.

The first is cardiorespiratory fitness and mortality, and it is genuinely impressive. The 2009 JAMA meta-analysis pooled 33 cohort studies — 102,980 participants, 6,910 deaths — and found a relative risk of 0.87 (95% CI 0.84 to 0.90) per 1-MET higher maximal aerobic capacity, with low-fitness participants at 1.70 times the mortality risk of high-fitness ones (95% CI 1.51 to 1.92) (JAMA 2009). A cohort of 122,007 patients referred for treadmill testing, followed a median 8.4 years with 13,637 deaths, found risk-adjusted mortality inversely proportional to fitness with no observed upper limit: elite versus low performers, adjusted hazard ratio 0.20 (95% CI 0.16 to 0.24), and elite versus merely high, 0.77 (95% CI 0.63 to 0.95). The mortality risk associated with low fitness there was comparable to or greater than that associated with coronary artery disease (1.29), smoking (1.41) or diabetes (1.40) (JAMA Netw Open 2018). An overview of meta-analyses spanning 199 cohort studies and more than 20.9 million observations reported a hazard ratio of 0.47 (95% CI 0.39 to 0.56) for high versus low fitness and an 11% to 17% reduction per 1-MET — while grading that evidence very low to moderate in certainty (Br J Sports Med 2024). Fitness also tracks with the risk associated with body weight: across 398,716 observations, overweight-fit and obese-fit groups did not differ significantly from normal-weight-fit for all-cause mortality (HR 0.96 and 1.11), while unfit groups at every weight carried two- to three-fold risks (Br J Sports Med 2025).

The second literature is physical activity dose and mortality. Harmonised accelerometer data from eight studies — 36,383 adults, mean age 62.6, followed a median 5.8 years — found hazard ratios across quarters of total activity of 1.00, 0.48, 0.34 and 0.27, with light activity alone showing 1.00, 0.60, 0.44 and 0.38, and moderate-to-vigorous activity 1.00, 0.64, 0.55 and 0.52 (BMJ 2019). The step-count meta-analysis found 7,000 steps a day, against 2,000, associated with a 47% lower risk of all-cause mortality (HR 0.53, 95% CI 0.46 to 0.60), a 38% lower risk of dementia and a 28% lower risk of falls, with inflection points around 5,000 to 7,000 steps (Lancet Public Health 2025). And the 94-cohort dose-response analysis estimated that if every insufficiently active adult reached 8.75 MET-hours a week, 15.7% of premature deaths would be averted (PMID 36854652).

Every one of those studies is observational. They measure fitness or activity at baseline and count deaths later, and they cannot separate the effect of training from the traits and circumstances that let a person train — the treadmill cohort, notably, was a clinical referral population. More to the point: none of them randomised anyone to an intensity, measured lactate, or assigned a zone. The activity literature's clearest intensity finding is that light activity carried its own mortality gradient — which is still an association, and still identifies no zone.

So the chain runs: fit people die later (observational); training raises fitness (randomised, and the increment is one to five mL·kg⁻¹·min⁻¹ in most trials); therefore training extends life (inferred, not demonstrated). Then a further step is added that has no support at all: that a specific sub-threshold intensity is the intensity that does it. Nothing in the mortality literature identifies an intensity. Nothing in the training literature shows sub-threshold work to be superior for the fitness variable that the mortality literature tracks. The leap from "fit people live longer" to "this zone extends life" is not supported by the evidence, and anyone making it is making it on vibes.

What zone 2 will not do

It will not, on the available evidence, raise your VO₂max faster than interval training does. The pooled head-to-head estimate in healthy adults aged 18 to 45 favours intervals by 1.2 mL·kg⁻¹·min⁻¹ (PMID 26243014), the advantage is larger in older adults (PMID 33836261) and present in cardiac patients (PMID 36753063), and in women the two were statistically indistinguishable (PMID 37084758). It has not been shown to produce a mitochondrial signalling response that harder training fails to produce; the PGC-1α meta-analysis found no significant difference between interval and continuous training, in a pooled estimate with substantial heterogeneity (PMID 40459444). Post-exercise fat oxidation was not higher after the uninterrupted moderate session than after sessions that included hard efforts, in 12 physically active young men (PMID 34637645). It will not substitute for volume: the recreational-runner trials that made polarized training look good were trials in which the polarized group trained substantially more (PMID 29952662). It will not compensate for skipped sessions: in the women's meta-analysis, women who completed more sessions improved more, in both formats (PMID 37084758). And it has never been randomised against anything with mortality as an endpoint, in anyone.

One more thing it will not do: guarantee you are safe from doing too much. In the live-high train-low camp, the group that did less low-intensity training and more strength and jump work was the group in which every athlete showed signs of overreaching, and whose performance got worse (PMID 30072913). That is at least a hypothesis worth holding — and a different one from the mitochondrial claim.

How to know it's working

The trials above did not measure zones. They measured outcomes, and those outcomes are what a person can reasonably track.

• Time over a fixed distance on a fixed course. The recreational-runner trials used a 10 km race on the same course before and after (PMID 23752040) and a 5 km time trial (PMID 29952662). Improvements in the range of 3.6% to 6.2% appeared over 10 to 12 weeks in both.
• Heart rate at a familiar pace. The submaximal variable is the one that moved most convincingly in the athlete data: oxygen uptake at the second ventilatory threshold rose 7.3 ± 5.8% in the skiers while VO₂max rose only 3.7% (PMID 30072913).
• VO₂peak, if you have access to a test. Trial-sized changes are modest and worth calibrating expectations against: 1.32 to 2.37 mL·kg⁻¹·min⁻¹ over eight weeks in cardiac rehabilitation (PMID 36753063), around 3.2 in the women's meta-analysis (PMID 37084758), 4.9 to 5.5 against non-exercising controls in young adults (PMID 26243014). Each 1-MET — roughly 3.5 mL·kg⁻¹·min⁻¹ — is the unit in which the mortality cohorts report their gradients (PMID 19454641, PMID 38599681) — but those cohorts measured fitness at baseline, not fitness gained, so a trained increment cannot be read back through that gradient.
• Blood pressure. In hypertensive participants both continuous and interval training lowered systolic pressure against controls, by 3.7 and 5.64 mmHg respectively (PMID 32125550).
• Body composition. Both training patterns reduced body fat by roughly 2.5% over 12 weeks in recreational runners, with no difference between them (PMID 29952662).
• Accumulated daily movement. Steps are a crude proxy, but they are the one with the best-characterised dose-response curve, with inflection points around 5,000 to 7,000 per day across mortality, cardiovascular disease, dementia and falls (PMID 40713949), and the light-intensity portion of that total carried its own mortality gradient in the accelerometer data (PMID 31434697).
• Whether you are still doing it in four months. Women who completed more sessions improved more, in both formats (PMID 37084758), and in 55- to 70-year-olds with prediabetes, the arm with no detectable decline in mitochondrial enzyme activity had completed 32 sessions across 16 weeks (PMID 39604207).

What none of this licenses is the audit at the bottom of the hill. Whether you crossed a line drawn from an estimated maximum heart rate for ninety seconds is not a variable in any trial described here, and the threshold methods those trials used disagree with each other by more than the width of the band you were defending (PMID 30761194, PMID 26849788). The defensible version of the zone-2 idea is duller than the marketed one: most aerobic training should be comfortable enough that you keep doing it, because a protocol you abandon produces nothing, and the talk test is a serviceable free way to check (PMID 25010379, PMID 18277826). The parts of the story about mitochondria, fat burning and lifespan are, for now, either unsupported or supported by evidence about something else.

Questions people actually ask

What should my zone 2 heart rate be? There is no number this article can hand you. A band computed from 220 minus your age comes off a population regression line, not off your physiology, and no study in the evidence base assembled here validated that equation; the displacement between it and your own threshold is unknown for you specifically. What that number is, is a rough population guess. What it is not, is a measurement.

Is there a zone 2 training calculator? Calculators exist and this piece does not endorse one, because the input they run on is the age-estimated maximum described above. The free alternative with validation behind it is the talk test: comfortable connected speech is likely possible below the ventilatory or lactate threshold and not likely above it, and the relationship held across walking, jogging, cycling, elliptical and stair stepper. Its known failure mode is erring toward passing when intensity has already exceeded the threshold.

Does zone 2 training work? Aerobic training works; that part is not in dispute. What the trials have not shown is that the sub-threshold intensity does something the other intensities cannot. Head to head for VO₂max, interval training was slightly ahead in healthy adults aged 18 to 45, ahead in older adults and in cardiac patients, and statistically indistinguishable from continuous training in women.

Does zone 2 training make you faster? In the two randomised trials of recreational runners here, the low-intensity-heavy groups did get faster — 5.0% over 10 km in one, 6.21% over 5 km in the other — but so did their comparison groups, by 3.6% and 5.49%, and neither between-group difference was statistically significant. In the 12-week trial the polarized group ran roughly 2.4 times the weekly running time to reach that equivalent result. Both trials were run in people who already ran.

Does zone 2 training burn fat, and is it good for fat loss? There is a moderate intensity at which the absolute rate of fat oxidation peaks during a session, and that is a real measurement. But in 12 physically active young men, post-exercise fat oxidation was 37.5% and 50% higher after moderate sessions interrupted by hard efforts than after the uninterrupted fat-max session. Over 12 weeks in 21 recreational runners, body fat fell 2.45% in the high-volume low-intensity group and 2.62% in the low-volume high-intensity group, with no meaningful difference between them.

Zone 2 training: how long, and is there a minimum time? No trial in this evidence set randomised session duration, so there is no minimum to report — only what the studies that produced these results asked people to do. The 12-week runner trial prescribed 295 ± 67 minutes a week across five running days and got 283; its comparison group did 117 minutes a week and matched it on race time. The older-adult meta-analysis found its favourable subgroups — for interval training, not for continuous work — at two sessions a week and around 40 minutes a session.

What is the zone 2 training myth? The zone 2 training myth is the mechanistic claim attached to the training, not the training itself: that sub-threshold work is uniquely or optimally mitochondrial. A 2025 narrative review in Sports Medicine examined that claim directly and concluded current evidence does not support zone 2 as the optimal intensity for improving mitochondrial or fatty-acid oxidative capacity, and that the recommendation derives largely from observational data on elite endurance athletes.

What is zone 2 training cycling, and is zone 2 training running any different? Zone 2 is defined by where your thresholds sit, not by the equipment, so the definition does not change between the bike and the road. What changes is the number: the heart rate identified at threshold depends on which analysis method was applied, and four methods compared in professional soccer players were judged not interchangeable.

What are the benefits of zone 2 training? The demonstrated benefits are the benefits of aerobic training generally. Against non-exercising controls, continuous endurance training raised VO₂max by 4.9 mL·kg⁻¹·min⁻¹ in healthy adults aged 18 to 45; in hypertensive participants it lowered systolic pressure by 3.7 mmHg; in recreational runners it reduced body fat by about 2.5% over 12 weeks; and in elite cross-country skiers, oxygen uptake at the second ventilatory threshold rose 7.3%. What is not demonstrated is a benefit belonging to the sub-threshold band specifically.

Does VO₂max decrease with age, and how does age affect VO₂max? That question belongs to the VO₂max after 50 piece linked below, which is where the evidence on aerobic decline is handled. The one age-related finding in this evidence set is from the women's meta-analysis, where the VO₂max response to training differed by age.

Does VO₂max predict longevity, and is a higher VO₂max better? As an association, yes: a 2009 JAMA meta-analysis of 33 cohorts found a relative risk of 0.87 per 1-MET higher capacity, and a cohort of 122,007 treadmill-tested patients found mortality inversely proportional to fitness with no observed upper limit. All of it is observational, none of it assigned anyone an intensity, and it measured fitness at baseline rather than fitness gained.

Related on Magellan: VO₂max after 50 · The 10,000-step threshold · What actually prevents falls.

Educational, not medical advice.