Default position: skeptical

The supplement industry is largely unregulated in the US — quality control varies wildly, label claims often outrun the evidence, and many products contain less of the active ingredient than they say. A 2015 New York Attorney General investigation found that 79% of store-brand herbal supplements tested didn’t contain meaningful amounts of the labeled herb.

Default position: don’t take a supplement unless you have a specific reason to. ‘Could help’ isn’t enough. The exceptions below are interventions where the evidence is strong enough to justify the trouble and cost for most healthy adults — or for clearly identified deficiencies.

The strong-evidence tier

These have the most robust evidence base and the cleanest case for supplementation in healthy adults.

• Vitamin D3 (1,000–2,000 IU/day, more if deficient). Roughly 40% of US adults are deficient. Easily corrected. Test your level (25-hydroxyvitamin D) before supplementing aggressively — target range is typically 30–60 ng/mL.
• Omega-3 (EPA + DHA) (1–2g combined/day). Most people undereat fatty fish. Algal oil works for vegetarians. Choose products with third-party testing for oxidation; rancid fish oil is worse than no fish oil.
• Creatine monohydrate (3–5g/day). The most thoroughly studied supplement on the market. Modest but real benefits for strength, lean mass, and increasingly, cognitive function in some populations. Cheap and safe.
• Magnesium glycinate or citrate (200–400 mg/day). Roughly half of Americans don’t meet RDA. Glycinate is better tolerated. Useful for sleep, recovery, general nervous system function.

The reasonable-evidence tier

Worth considering for specific contexts, but evidence is more situational.

• Vitamin B12 — if vegan/vegetarian or over 50 (where absorption declines). Methylcobalamin form, 500–1,000 mcg several times a week is plenty.
• Iron — only if you’ve tested low (ferritin <30). Don't supplement without testing; excess iron is harmful.
• Zinc (10–15 mg) — useful in acute illness; doesn’t need chronic supplementation if your diet includes meat, shellfish, legumes, or pumpkin seeds.
• Whey or plant protein — useful if you can’t hit ~1.6–2.0 g/kg/day of protein through food, which is most people.
• Caffeine — performance and cognitive benefits at 3–6 mg/kg. Most reliable supplement on the market by far. Calibrate to half-life.

The thin-evidence tier

Lots of marketing, much weaker science. Useful sometimes, often not. Don’t build a stack around these.

• Probiotics for general ‘gut health’ — strain-specific evidence exists; broad multi-strain supplementation is mostly hopeful.
• Collagen — anecdotally helpful for some; weak evidence for skin/joint outcomes beyond what protein generally provides.
• Adaptogens (ashwagandha, rhodiola) — modest effects on stress markers in some studies; not transformative.
• Multivitamins — most studies find no all-cause mortality benefit in healthy adults; modest cognitive benefit in some older-adult cohorts. Better to identify specific shortfalls and supplement them.

The not-worth-it tier

Persistent marketing, persistently disappointing evidence.

• Most fat-burning supplements (caffeine + green tea extract is the only thing with even modest support).
• Branched-chain amino acids (BCAAs) for muscle growth — leucine alone or whole protein outperforms.
• Glucosamine/chondroitin for joint pain — repeatedly underwhelming results.
• Detox supplements — your liver and kidneys handle this without help.

Suggested starter approach

1. Get bloodwork. Vitamin D, B12, ferritin, magnesium (RBC), omega-3 index. Supplement what you’re actually low in.
2. Add creatine if you train. The risk-benefit math is excellent.
3. Add omega-3 if you don’t eat fatty fish 2–3x per week.
4. Magnesium glycinate before bed if you have any sleep or recovery issues.
5. Resist the urge to stack. More pills isn’t better health.

Most of what people try to fix with supplements is better addressed by diet, sleep, movement, and stress management. If those four aren’t dialed in, no supplement stack will compensate.