TL;DR
The longevity research from the last 80 years agrees on a small set of things: don't smoke, move daily, sleep 7-9 hours, eat mostly plants + fish, drink little alcohol, maintain relationships, manage stress. That's it. The supplements, biohacks, ice baths, and red light therapy that dominate longevity content have effect sizes ranging from "tiny" to "actively harmful". This article shows what the actual data looks like, ranked by effect size.

Longevity has become a content category. Books, podcasts, supplements, $5K/year clinics that test your epigenetic age. Andrew Huberman, Peter Attia, Bryan Johnson, and 50 influencers selling NMN.

The frustrating thing for newcomers is that the actual research is much simpler — and much less marketable — than the content stack suggests. The boring stuff has 100× the evidence of the exciting stuff. This article goes through what the data actually shows.

The longevity equation, simplified

Your maximum lifespan has roughly three layers:

  1. Genetics — accounts for ~20-25% of lifespan variance (twin studies, Herskind 1996, ongoing)
  2. Lifestyle — accounts for ~50-60% (diet, exercise, sleep, smoking, alcohol, social)
  3. Environment + luck — accidents, infections, healthcare access, pollution: ~15-25%

The interesting bucket is lifestyle. It's modifiable, and the effect sizes are enormous. Comparing top-quartile lifestyle to bottom-quartile across multiple studies: 12-14 years longer life expectancy. That's bigger than most genetic mutations.

Want to estimate where you currently are? The biological age calculator incorporates these factors into a lifestyle-based bio-age estimate.

What actually works, ranked by effect size

1. Don't smoke (effect: -10 years if you do, +10 if you quit)

Smoking is by far the largest modifiable longevity factor. Lifelong smoker vs never-smoker: 10 years of life expectancy lost on average (Doll et al. 2004, Pirie 2012, ongoing meta-analyses). The good news: quitting works fast. By year 1, cardiovascular risk drops 50%. By year 10, lung cancer risk drops 50%. By year 15, all-cause mortality is similar to never-smokers.

There is no other lifestyle intervention with this magnitude of effect. Nothing else competes. If you smoke, the single highest-leverage longevity move is quitting — full stop.

2. Move daily (effect: -2 to -4 years for sedentary, +3 years for active)

Not "exercise" — movement. The Lancet 2023 meta-analysis (Paluch et al.) is the cleanest single data point: 8,000 steps/day = 51% lower all-cause mortality vs 4,000 steps. The dose-response is monotonic up to ~10K, then flattens.

Specific findings:

Track your actual movement: the walking distance lifetime calculator shows your cumulative km and the years-gained from your current pattern.

⚡ Truth Series take
The "10,000 steps" myth originated from a 1965 Japanese pedometer marketing campaign — not science. Actual research suggests 7,000-9,000 is the optimal mortality-reduction zone. More is fine, not magical.

3. Sleep 7-9 hours consistently (effect: ±2 years)

Sleep is where most people are silently underperforming. The U-shaped mortality curve is well-replicated: <6 hours and >10 hours both increase risk; 7-9 is the sweet spot.

Specific findings:

Most adults need 7-9 hours; some genetically need closer to 9. The "I only need 5 hours" crowd is real but rare (~1% of population). Most "short sleepers" are accumulating chronic deficit and rating themselves "fine" because they've forgotten what real rest feels like.

The real sleep-need calculator adjusts for your chronotype, age, exercise, and stress to estimate your actual need vs current pattern.

4. Eat mostly plants + fish, less ultra-processed (effect: ±3 years)

The Mediterranean diet has the strongest meta-analytic evidence of any specific dietary pattern. PREDIMED (2018) randomized 7,447 high-risk adults to Mediterranean vs control: 30% fewer cardiovascular events over 4.8 years. EPIC + Nurses Health Studies show similar long-term mortality benefit.

The principles that replicate:

What's emerging as the worst dietary factor: ultra-processed food (UPF) — things with industrial-only ingredients (HFCS, hydrogenated oils, modified starches, flavor enhancers). NOVA classification. Recent meta-analyses associate >50% of calories from UPF with 25-30% higher all-cause mortality (Bonaccio 2022, Schnabel 2019).

5. Drink little alcohol — ideally zero (effect: -1 to -3 years moderate, -5+ years heavy)

The most-revised longevity finding of the past decade. The "1 glass of wine/day is healthy" claim came from observational studies that compared drinkers to abstainers — but the abstainer group included sick people who quit drinking because of illness. This "sick quitter effect" made any drinking look protective.

The Lancet 2018 GBD analysis (28M person-years) and 2024 update settle it: the safe level of alcohol consumption for health is zero. Risk increases monotonically with intake. Even 1 drink/day raises cancer risk slightly. The "J-curve" was a statistical artifact.

This doesn't mean you must be a teetotaler. It means: if you're optimizing for longevity, less is better than more, and zero is better than less. The AUDIT-C self-screen can identify whether your drinking is in the zone where reduction would matter.

6. Stay socially connected (effect: ±2-3 years)

Loneliness is one of the most underrated longevity factors. Holt-Lunstad's meta-analyses (2010, 2015) found social isolation comparable to smoking 15 cigarettes/day in mortality impact. Dunbar, Pinker, and Blue Zones research all point to the same: dense social ties extend life.

What replicates:

Modern life has industrialized loneliness. Track relationship maintenance proactively — the friendship decay calculator shows how relationships you don't actively tend to drift faster than people realize.

7. Manage chronic stress (effect: ±1-2 years)

Chronic stress shortens telomeres (Epel 2004), elevates baseline cortisol, accelerates cardiovascular aging, suppresses immune function. The Whitehall II studies (1985-ongoing) showed civil servants in low-control jobs had 50% higher CVD mortality than high-control peers — independent of income.

The interventions with strongest evidence:

The meditation hours lifetime tool shows the cumulative dose effect — long-term meditators show measurable brain change (Davidson lab research) in ways short-term practice doesn't.

What's NOT supported by strong evidence (yet)

The supplement industry, biohackers, and longevity podcasts will hate this section. But the evidence is what it is.

NMN, NR (NAD+ precursors)

Theoretical mechanism (NAD+ declines with age, supplementing might help). RCT evidence in humans: small effects on some biomarkers, no demonstrated mortality or healthspan extension. Bryan Johnson takes them. So do thousands of biohackers. The mortality data isn't there. Could be promising, isn't proven.

Resveratrol

Got popular in the 2000s based on Sinclair's mouse studies. Two decades later, human trials have been disappointing. Sirtris Pharmaceuticals (acquired by GSK for $720M) shut down without a viable drug. Take a multi-vitamin if you want; the evidence isn't there for longevity.

Red light therapy

Emerging evidence for skin healing, mitochondrial function in muscle. Not yet evidence for lifespan extension. Effect sizes (where measured) are small. Wait 5 years before treating it as longevity intervention.

Cold plunges / ice baths

Effects on inflammation, mood, alertness — yes, some. Effect on lifespan — no human data. Plenty of anecdotes; no longitudinal mortality studies.

Most peptides

BPC-157, TB-500, etc. — primarily small-animal data + anecdotal human use. Not approved for human longevity use. Potentially risky long-term.

Intermittent fasting

Better evidence than most biohacks. Some metabolic benefits in RCTs (improved insulin sensitivity, weight loss). Mortality benefit not yet demonstrated in humans (large studies underway). Probably positive, but treat the certainty as moderate, not high.

Cryotherapy, hyperbaric oxygen, ozone therapy

Marketing has outpaced evidence. Skip until human longevity data emerges.

The Blue Zones — what they actually share

Dan Buettner's Blue Zones research identified 5 regions with unusual centenarian density: Sardinia (Italy), Okinawa (Japan), Nicoya (Costa Rica), Ikaria (Greece), Loma Linda (US, Seventh-Day Adventists).

The methodology has been criticized (some Sardinian centenarian records are genealogically suspect). But the lifestyle pattern across regions is consistent and matches everything above:

Notice: zero supplements, zero gadgets, zero podcasts. The interventions that produce the longest-lived populations are structural (community, food culture, walkable geography) and lifestyle (movement, sleep, relationships).

The "longevity escape velocity" question

A subset of longevity researchers (Aubrey de Grey, David Sinclair) believe radical lifespan extension (150+ years) is achievable in coming decades through cellular reprogramming, senolytics, and AI-driven drug discovery. The evidence ranges from "promising in mice" to "completely speculative in humans".

Honest take for non-researchers: the speculative future doesn't change what you should do now. If escape velocity arrives in 30 years, the people positioned to benefit are those still alive and metabolically healthy. The boring lifestyle stack is the bridge to whatever comes next.

The 90-day longevity reset

If you want to operationalize this, here's a 90-day protocol with research-backed leverage in roughly the order of effect size:

Days 1-30: Move + sleep

Days 31-60: Add structure

Days 61-90: Optimize

Take the composite score at the start and end. Most people see meaningful biomarker improvements by day 60-90.

FAQ

Does fasting / intermittent fasting work?

Probably yes, modestly. The mechanisms (autophagy, insulin sensitivity) are real. Human mortality data is still emerging. Time-restricted eating (TRE) — 8-10 hour eating windows — is the gentlest version with reasonable evidence. Multi-day fasts have benefits in metabolically-healthy adults; risks in some populations. Not magical, but reasonable.

What about strength training for women?

Critical, and consistently under-emphasized. Bone density loss accelerates after perimenopause; strength training is the most effective preventive intervention against osteoporosis. Recent meta-analyses (Liu 2024) show resistance training reduces all-cause mortality 23% in women, similar to men.

Is coffee good or bad?

Net positive for most adults, in moderation (3-4 cups/day max). Meta-analyses show 7-15% reduced all-cause mortality at moderate coffee intake. Mechanisms: antioxidants, cardiovascular effects, modest reduction in liver cancer + Type 2 diabetes risk. Caveats: pregnant women should limit; some genetic slow metabolizers may not benefit.

Should I get my biological age tested?

If you have $200-400 to spare, services like TruDiagnostic, Elysium Index, or similar offer epigenetic-clock testing (Horvath, GrimAge, DunedinPACE). Useful as a baseline + measuring response to interventions. Not necessary — the lifestyle factors that improve bio-age are knowable without testing. But helpful for the data-driven.

What's the most overrated longevity advice?

"Eat clean." It's vague and often becomes orthorexia. "Take collagen." Effect sizes are tiny. "Optimize your microbiome." Promising but premature for most consumer interventions. "Cold plunges every morning." Trendy, no human longevity data. "Track everything." Often creates anxiety that itself accelerates aging.

How much of longevity is luck?

More than you'd like. Cancer, accidents, and infections still kill people who did everything right. The lifestyle factors push the median outcome 10+ years; they don't guarantee individual outcomes. The framing that helps: you're playing a probability game, not a deterministic one.

The honest closing

The longevity research is unsexy and stable: don't smoke, move, sleep, eat plants + fish, drink little, stay connected, manage stress. That's the ~80% of the effect size. The other 20% is fragmented across hundreds of smaller interventions of varying evidence quality.

The supplement industry, biohacker influencers, and longevity clinics are largely selling the 1-5% optimization layer at high prices to people who haven't yet implemented the 80% layer. The 80% layer is free. It just requires consistent action over decades.

If you want a baseline, take the composite score — it scores your current pattern across 10 dimensions and shows the highest- leverage moves. Free, no sign-up, runs in your browser.

Sources