Chocolate and cocoa: the big trials tested a supplement

The largest and cleanest evidence here was never about chocolate. COSMOS randomised older adults to a cocoa extract supplement, and over a median 3.4 years it did not stop them developing high blood pressure, at a hazard ratio of 0.96. A reanalysis of that same trial shows how fragile these readings are. One statistical model returned a small significant benefit on a composite outcome, and a standard Cox regression of the identical composite returned nothing. Pooled across 31 randomised trials, cocoa lowered total cholesterol by about 8 mg/dL, LDL by about 9, fasting glucose by about 5, systolic pressure by 2.52 mmHg and diastolic by 1.58, and did nothing to body weight, waist or HbA1c. Nine of those 31 trials were at low risk of bias and the other 22 raised concerns, and every pooled outcome was graded moderate quality after a downgrade for imprecision. None of it arrived in a bar of chocolate.

The numbers, per 100 g

What your body absorbs

• Cocoa's own theobromine raises how much of its flavanols reaches the blood, and the boost needs a working colon. — 10 volunteers with a functional colon and 10 volunteers with an ileostomy (n = 20) Evidence B source
  Co-ingesting flavan-3-ols with cocoa methylxanthines, mainly theobromine, raised peak circulating levels of flavan-3-ol metabolites by 21 plus or minus 8% (p < 0.05), while caffeine gave 10 plus or minus 8% and was not significant; in ileostomists cocoa methylxanthines did not change circulating metabolite levels, and ileal fluid metabolites were unchanged

What it does to your health

• Pooled trials of cocoa move cholesterol, glucose and blood pressure a little, and body weight not at all. — adults with and without comorbidities across 31 randomised trials (n = 1986) Evidence A source
  No effect on body weight, body mass index, waist circumference, triglycerides, HDL cholesterol or HbA1c; reductions in total cholesterol of 8.35 mg/dL (95% CI -14.01 to -2.69), LDL of 9.47 mg/dL (-13.75 to -5.20), fasting glucose of 4.91 mg/dL (-8.29 to -1.52), systolic pressure of 2.52 mmHg (-4.17 to -0.88) and diastolic of 1.58 mmHg (-2.54 to -0.62)

What people believe that the data does not support

• The review that looked at ten years of diet and acne research put chocolate on the promoting side, not the exonerated one. — 53 articles meeting eligibility, comprising 7 randomised controlled trials, 4 uncontrolled open label studies, 17 case-control, 22 cross-sectional and 3 descriptive studies (n = 53 studies) Evidence C source
  Acne-promoting factors identified were high glycaemic index and load foods, dairy products, fatty food and chocolate, while acne-protective factors were fatty acids and fruit and vegetable intake; the review states that the role of specific components within chocolate, cocoa against dark against milk, remains an unsolved issue for future research
• Europe refused the claim that cocoa flavanols keep blood pressure normal. — — Evidence E source
  The register entry for cocoa flavanols and maintenance and promotion of a normal blood pressure carries the status Non-authorised; across the seven register entries naming cocoa or cocoa flavanols in this snapshot, authorised claims number zero and refusals seven
• The one cocoa claim America does allow says in its own wording that the evidence is very limited. — — Evidence E source
  FDA said it intends to exercise enforcement discretion for wordings that pair the claim with a statement that the agency has concluded there is very limited scientific evidence for it, with the qualifying product defined as high flavanol cocoa powder containing at least 4 per cent naturally conserved cocoa flavanols

Who this works differently for

• Over a median 3.4 years, cocoa extract did not stop older adults developing high blood pressure. — COSMOS participants free of hypertension at baseline, women aged 65 and over and men aged 60 and over, mean age 71.1 years, 59% women (n = 8905) Evidence A source
  Over a median 3.4 years there was no effect on incident hypertension, incidence rates 7.1 against 7.4 per 100 person-years, hazard ratio 0.96 (95% CI 0.88 to 1.05); in subgroups, those starting with systolic pressure below 120 mm Hg had a hazard ratio of 0.76 (0.64 to 0.90) while those at 120 to 139 had 1.05 (0.93 to 1.18), interaction p = 0.002
• In a pilot of 35 people against 11 controls, 40 g of dark chocolate three times a week during dialysis raised serum magnesium without moving potassium or phosphorus, which is a reason to ask a nephrologist rather than to change a renal diet. — patients with chronic kidney disease on haemodialysis, 35 completing in the chocolate group and 11 in the control group (n = 59 enrolled, 46 completed) Evidence C source
  Serum magnesium rose significantly in the dark chocolate group from a median 2.5 (2.4 to 2.6) to 2.7 (2.6 to 2.8) mg/dL (P < .01), with no change in plasma potassium or phosphorus

What is still unknown

• The same cocoa trial reads as a null or as a win depending on which statistical model is used. — the full COSMOS population of healthy older US adults, women aged 65 and over and men aged 60 and over, followed a median of 3.6 years (n = 21442) Evidence B source
  Win ratio analysis gave cocoa wins of 3.41% against placebo wins of 2.87%, a reciprocal win ratio of 0.84 (95% CI 0.72 to 0.99) and net treatment benefit of -0.54% (-1.04 to -0.03), p = 0.037, while Cox regression of the identical composite gave a non-significant hazard ratio of 0.90 (95% CI 0.79 to 1.03)
• Fewer than a third of the cocoa trials behind those numbers were at low risk of bias. — the 31 randomised controlled trials pooled in that meta-analysis (n = 1986) Evidence A source
  Nine of the 31 trials showed a low risk of bias, while the remaining 22 showed some concerns, mostly for inconsistent reporting of outcomes; every pooled outcome including blood pressure, lipids and glycaemia was graded moderate quality, downgraded one point for imprecision because of differing samples and wide confidence intervals

The bottom line

Eat chocolate because you want chocolate. The one claim the United States allows says as much in its own wording: cocoa flavanols in high flavanol cocoa powder may reduce the risk of cardiovascular disease, although the agency has concluded there is very limited scientific evidence for the claim. The qualifying product is a powder holding at least 4 percent naturally conserved cocoa flavanols. Europe went the other way and refused all seven cocoa entries in its register, the blood pressure one included. A bar of 70 to 85 percent dark chocolate runs 598 kcal per 100 g, and that is the number to weigh against a 2.52 mmHg change in systolic pressure measured on capsules. Two smaller findings are worth carrying. Cocoa brings its own theobromine, and that raised peak blood levels of flavanol metabolites by 21 percent in people with a working colon and not at all in people with an ileostomy, so the gut is part of the mechanism. And in a pilot during haemodialysis, 40 g of dark chocolate raised serum magnesium without moving potassium or phosphorus, in 35 people against 11 controls. That is a pilot, and if you are on dialysis your food list is a clinical decision rather than a reading of this page. If you have acne, the review covering ten years of diet and acne research placed chocolate among the promoting factors, and said the question of cocoa against dark against milk chocolate is still unsolved.

Sources

1. cho-dr-05 · randomised controlled cross-over bioavailability study
   In a randomized, controlled, triple cross-over study in volunteers with a functional colon (n = 10), co-ingestion of flavan-3-ols and cocoa methylxanthines, mainly represented by theobromine, increased peak circulatory levels (Cmax) of flavan-3-ols metabolites (+21 ± 8%; p < 0.05).
   Free Radic Biol Med, 2023 · checked 2026-09-22
2. cho-dr-03 · systematic review and meta-analysis of randomised controlled trials
   Yet, there was a reduction in total cholesterol (-8.35 mg/dL, 95% CI -14.01; -2.69 mg/dL), LDL-c (-9.47 mg/dL, 95% CI -13.75; -5.20 mg/dL), fasting blood glucose (-4.91 mg/dL, 95% CI -8.29; -1.52 mg/dL), SBP (-2.52 mmHg, 95% CI -4.17; -0.88 mmHg), and DBP (-1.58 mmHg, 95% CI -2.54; -0.62 mmHg).
   Nutrients, 2024 · checked 2026-09-22
3. cho-dr-06 · systematic review
   Acne-promoting factors include high GI/GL food, dairy products, fat food, and chocolate, whereas acne-protective factors include fatty acids, fruit, and vegetable intake.
   Int J Dermatol, 2021 · checked 2026-09-22
4. cho-dr-08 · regulatory register entry
   POL-HC-6949 Cocoa flavanols Maintenance and promotion of a normal blood pressure Non-authorised
   EU Register on nutrition and health claims, заява POL-HC-6949, знімок 2026-09-21 · checked 2026-09-22
5. cho-dr-09 · regulatory decision
   Cocoa flavanols in high flavanol cocoa powder may reduce the risk of cardiovascular disease, although the FDA has concluded that there is very limited scientific evidence for this claim.
   FDA, HFP Constituent Update, 1 лютого 2023, кваліфікована заява про флаваноли какао · checked 2026-09-22
6. cho-dr-01 · randomised double-blind placebo-controlled trial
   Over a median follow-up of 3.4 years, cocoa extract supplementation had no significant effect on incident hypertension in an intention-to-treat analysis, with incidence rates of 7.1 and 7.4 per 100 person-years in cocoa and placebo groups, respectively (hazard ratio, 0.96 [95% CI, 0.88-1.05]).
   Hypertension, 2025 · checked 2026-09-22
7. cho-dr-07 · controlled pilot study, secondary analysis
   After 2 months of supplementation, serum levels of Mg2+ increased significantly in the dark chocolate group from 2.5 (2.4-2.6) to 2.7 (2.6-2.8) mg/dL (P < .01).
   J Ren Nutr, 2025 · checked 2026-09-22
8. cho-dr-02 · secondary analysis of a randomised controlled trial
   GPC analyses showed cocoa flavanol wins of 3.41% and placebo wins of 2.87%, yielding a reciprocal WR of 0.84 (95% CI 0.72-0.99) and negative NTBs of - 0.54% (- 1.04 to - 0.03), p = 0.037.
   Eur J Epidemiol, 2026 · checked 2026-09-22
9. cho-dr-04 · risk of bias and GRADE assessment within a meta-analysis
   Of the 31 studies included in this meta-analysis, nine showed a low risk of bias [17,23,34,40,52,60,61,62,63], as they met the RoB 2 criteria, including prior registration of studies on clinical trial registry platforms.
   Nutrients, 2024 · checked 2026-09-22