If a pharmaceutical company developed a single agent that produced the all-cause mortality reduction associated with regular exercise, it would be the most successful drug in history. The cohort signal across millions of adults is steeper, more consistent, and more biologically plausible than any other longevity intervention currently on offer. This page lays out what the data actually shows: the dose-response, the difference between aerobic and resistance modalities, the minimum effective dose, and the gap between observational signals and what RCTs have shown on intermediate endpoints.
The cardiorespiratory fitness signal
Mandsager 2018 (JAMA Network Open) followed 122,007 adults through treadmill stress testing and tracked all-cause mortality over a median 8.4 years ( Mandsager et al. 2018, n=122007 ). The dose-response was monotonic across the entire fitness distribution. Each additional MET of peak fitness associated with about 12% lower all-cause mortality. The hazard ratio comparing the lowest fitness quintile to the highest was 5.04, larger than the hazard ratios for smoking, hypertension, and type 2 diabetes combined.
A few features of this signal are worth noting. First, the curve does not plateau in healthy ranges. Even at the elite end (above 14 METs in adults under 50), additional fitness was associated with further mortality reduction. Second, the relative gain from moving from "sedentary" to "below average" is comparable to the gain from moving from "above average" to "elite". The marginal value of the first MET is enormous; the marginal value of the tenth is still positive but smaller. Third, the association persisted after adjustment for smoking, BMI, hypertension, diabetes, and lipids, which are the obvious confounders.
Cardiorespiratory fitness in this dataset corresponded roughly to VO2 max ranges of 25 to 60 mL/kg/min in middle-aged adults. The 1 MET difference is approximately 3.5 mL/kg/min, achievable by moving from "no aerobic training" to "20 to 30 minutes of moderate aerobic activity 3 to 4 times per week" over 8 to 12 weeks in untrained adults.
The walking dose-response
Step count cohort data extends the fitness signal to ambulatory activity that does not require structured training. Saint-Maurice 2020 (JAMA, n=4,840 US adults, 10-year follow-up) reported a steep dose-response from 4,000 to about 10,000 steps per day, with hazard ratios falling from 1.0 to about 0.49 ( Saint-Maurice et al. 2020, n=4840 ). Step intensity (cadence) added little after step count was controlled, suggesting accumulated volume matters more than how briskly any particular block is walked.
Paluch 2022 pooled 15 international cohorts (n=47,471) and refined the picture ( Paluch et al. 2022, n=47471 ). Adults under 60 hit diminishing returns around 8,000 to 10,000 steps per day. Adults 60 and over plateaued earlier, around 6,000 to 8,000 steps per day. The "10,000 steps" figure that became a marketing number in the 1960s pedometer launch turns out to be roughly correct for the under-60 group and roughly 30% too high for older adults.
The practical read: any movement from sedentary baseline (typically 3,000 to 5,000 daily steps in office workers) toward 7,000 captures most of the cohort-level mortality reduction. The next 3,000 steps add smaller marginal benefit. Above about 12,000 daily steps, the data is sparse and the signal flat to slightly inverted in some cohorts, possibly reflecting confounding by chronic illness driving high activity in retirees.
Resistance training and the strength-mortality association
Aerobic training dominates the longevity literature for historical reasons (treadmills and accelerometers were the first available outcome instruments), but resistance training has its own independent mortality signal. Multiple cohort studies suggest 2 to 3 sessions per week of resistance work associates with 15 to 20% lower all-cause mortality after adjustment for aerobic activity. The effect is independent of and additive to the aerobic dose.
The mechanism is plausible. Skeletal muscle is the largest endocrine organ by mass, the dominant site of insulin-stimulated glucose disposal, and the substrate for grip strength and gait speed (both of which independently predict mortality in elderly populations). Sarcopenia, the age-related loss of about 1% of muscle mass per year after age 50, is one of the strongest predictors of late-life functional decline.
Morton 2018 (BJSM meta-analysis, n=1,863 across 49 RCTs) anchored the protein-and-resistance-training dose-response ( Morton et al. 2017, n=1863 ). Hypertrophy gains scaled with protein intake up to about 1.6 g/kg/day, with diminishing returns beyond that. Strength gains tracked hypertrophy but with a more modest dose-response. Schoenfeld 2017 reviewed weekly volume and found 10 or more sets per muscle per week was associated with larger hypertrophy gains, with the plateau unclear ( Schoenfeld et al. 2016 ).
For longevity-oriented training (rather than maximum hypertrophy), the dose can be much lower. Paoli 2024 reviewed minimum effective dose data and concluded 30 to 60 minutes per week of resistance training, distributed across 2 sessions, preserves most strength and lean mass gains compared to higher volumes ( Paoli et al. 2024 ). This is the dose that should be the floor for any adult over 40 trying to age well.
The minimum effective dose
Three numbers anchor the practical floor:
- 150 minutes of moderate aerobic activity per week, or 75 minutes of vigorous activity per week, drops all-cause mortality risk roughly 20 to 30% versus complete inactivity. This is the WHO and US Department of Health threshold.
- 2 sessions of resistance training per week of 30 to 45 minutes each, hitting major muscle groups, captures most of the lean mass and strength preservation benefit. 4 sessions per week add about 30 to 50% more hypertrophy in trained populations but the marginal mortality return is small.
- 7,000 daily steps captures most of the step-count mortality benefit in adults under 60.
These are floors, not ceilings. The dose-response continues above each threshold, but the slope flattens. For most adults, the binding constraint is consistency (does the program survive a busy quarter at work?) rather than peak intensity. A program that gets done 50 weeks per year at the floor dose dominates a program that runs 12 weeks per year at twice the floor.
Aerobic intensity: zone 2, threshold, and VO2 max work
Within aerobic training, intensity distribution matters. The dominant framework treats training as a 3-zone system: zone 1 (easy, conversational, around 60 to 75% of max heart rate), zone 2 (moderate, comfortable but not conversational, around 70 to 85%), and zone 5 (max, intervals at 90% or higher). Endurance athletes typically distribute volume around 70 to 80% zone 1, 5 to 10% zone 2, and 10 to 20% zone 5.
For non-athletes pursuing longevity, the practical recommendation is roughly 80% of aerobic time in zone 1 to 2 (steady-state, easy to moderate) and 20% in higher-intensity work (intervals or threshold sessions). This 80/20 distribution captures most of the VO2 max gain while keeping accumulated stress low enough for adherence over years. Pure interval training produces faster VO2 max gains over 8 to 12 weeks but risks injury and burnout when sustained. Pure zone 1 training produces slower VO2 max gains but is sustainable indefinitely.
VO2 max specifically responds well to 1 to 2 sessions per week of high-intensity intervals, typically 4 by 4 minute intervals at 90 to 95% of max heart rate with 3 minute recoveries. A meta-analysis of HIIT versus moderate continuous training in middle-aged adults found roughly 1.5 to 2.5 mL/kg/min greater VO2 max gain per 8-week training block, a clinically meaningful effect.
Why exercise stacks with everything else
Exercise is the only intervention with a documented effect across essentially every longevity-relevant axis: cardiorespiratory fitness, muscle mass, insulin sensitivity, BP, lipids, mood, sleep architecture, cognitive function, bone density, immune function, and inflammatory tone. This is unusual. Most pharmacological interventions move 1 to 3 axes; exercise moves 10 or more.
The breadth has two implications. First, exercise is uniquely difficult to substitute. No combination of supplements or pharmaceuticals reliably replicates the full multi-system effect; the partial substitutes (semaglutide for weight, statins for LDL, metformin for glucose) move single axes and often compromise others (semaglutide accelerates lean mass loss without resistance training, metformin attenuates aerobic adaptation). Second, the foundational nature of exercise means it should be the floor before adding other interventions, not a supplement to them. An adult exercising 0 hours per week and considering rapamycin, NAD precursors, and senolytics is mis-prioritizing.
Konopka 2018 illustrated this point in a metformin trial ( Konopka et al. 2018, n=53 ). Older adults randomized to metformin during 12 weeks of aerobic training had attenuated VO2 max gains and attenuated insulin-sensitivity gains compared to placebo. The mitochondrial AMPK signaling that exercise drives appears partially blocked by metformin co-administration. The implication is not that metformin is harmful; it is that the layering order matters, and exercise should be the foundation.
Healthspan vs lifespan: the curve flattens
The cohort data on lifespan extension from exercise is impressive but the more important finding may be on healthspan. The compression-of-morbidity argument holds that exercise does not just add years to life; it shifts the onset of disability and chronic disease later, compressing the years of poor function into a smaller window before death. Functional capacity at age 70 in active adults often resembles functional capacity at age 50 in sedentary adults.
This is why grip strength, gait speed, and chair-rise test performance are increasingly used as healthspan biomarkers. They predict not just mortality but functional independence. An adult who can rise from a chair without using their hands, climb 4 flights of stairs without stopping, and grip 30 kg or more is in the high-function tail of their age distribution, with strong correlations to remaining healthy life.
What exercise does not do
A few claims persistently appear in fitness marketing and are not well-supported:
- "Exercise burns the most calories of any intervention." A 60 minute moderate session burns 300 to 500 kcal in most adults. Multi-day caloric deficit math means exercise is a modest contributor to weight loss compared to dietary intake; the strongest fat-loss leverage point is consistent calorie modulation, not exercise volume. Exercise is essential for body composition (lean mass preservation during deficit) but not the primary lever for total weight.
- "Cardio destroys gains." Concurrent training (aerobic + resistance) attenuates hypertrophy slightly versus resistance-only protocols, but the magnitude is small (roughly 5 to 10% lower hypertrophy in concurrent vs resistance-only) and only emerges at high aerobic volumes. For most adults pursuing health rather than peak hypertrophy, the interference effect is irrelevant.
- "Exercise raises testosterone substantially." Acute resistance sessions transiently raise testosterone for 30 to 90 minutes post-session; chronic baseline testosterone shifts are small (5 to 15%) and tend to disappear when adjusted for body composition changes. Exercise does not cure low testosterone in clinically hypogonadal men.
Practical synthesis
The evidence-anchored prescription for an adult pursuing exercise as a longevity intervention:
- Base aerobic volume. 150 to 300 minutes per week of zone 1 to 2 work (walking, easy cycling, easy running). This is the floor that captures most of the mortality reduction.
- VO2 max sessions. 1 to 2 sessions per week of 4 by 4 minute intervals or equivalent. 8 to 12 weeks to detect VO2 max gain on a treadmill or bike test.
- Resistance training. 2 to 3 sessions per week, 30 to 60 minutes each, full-body or upper/lower split. Compound lifts (squat, deadlift, press, pull) at 6 to 12 reps for 3 to 5 sets per exercise.
- Daily ambulation. 7,000 to 10,000 steps. Distribute across the day rather than concentrated in single sessions.
- Recovery. 1 full rest day per week minimum. Sleep 7 to 9 hours. Schedule cold and sauna away from resistance sessions if used.
A program that hits these targets for 50 weeks per year produces fitness, lean mass, and metabolic outcomes in the upper tail of the age distribution. This is the foundation everything else stacks on top of.
