A single square of dark chocolate broken from a bar on a white ceramic plate, beside a small heap of cacao nibs and one whole roasted cacao bean

Dark Chocolate and Your Health: What the Studies Say and Don't

The pitch for dark chocolate arrives as a headline every few months: good for the heart, good for the brain, good for blood sugar. Every few months there is a counter-headline about lead. Both come from real studies. Neither, on its own, tells you what an ounce of a 70 percent bar does. This page grades the claims one at a time, names the study behind each, and separates what a supplement did in a trial from what a bar does on a plate. Our craft chocolate guide covers what a good bar is; this page covers whether it is doing anything for you.

What "good for you" would have to mean

Nearly every health claim about chocolate runs through one family of compounds: cocoa flavanols, of which epicatechin is the best studied. The claims rest on two kinds of evidence. Randomized trials give people a measured dose of flavanols, usually as a cocoa extract capsule or a specially processed high-flavanol drink, and compare them with a placebo. Cohort studies ask large groups what they eat and watch what happens to them over decades. The first kind can show cause. The second can only show that people who eat chocolate differ from people who do not. Keep that distinction in hand, because most chocolate headlines are built on the second kind and written as if they were the first.

The one claim regulators accept

In 2012 the European Food Safety Authority reviewed the flavanol trials and concluded that a cause-and-effect relationship had been established between cocoa flavanols and the maintenance of normal endothelium-dependent vasodilation, which is the ability of blood vessels to relax and widen. EFSA's proposed wording was modest: cocoa flavanols "help maintain endothelium-dependent vasodilation, which contributes to normal blood flow." The wording the European Commission actually authorized in 2013 is plainer still: cocoa flavanols "help maintain the elasticity of blood vessels, which contributes to normal blood flow." The condition is 200 mg of cocoa flavanols a day, which EFSA said could come from 2.5 g of high-flavanol cocoa powder or 10 g of high-flavanol dark chocolate. The words "high-flavanol" are doing real work there, and we come back to them below.

The related claim, that cocoa lowers blood pressure, holds up but is small. A 2017 Cochrane review pooled 35 randomized trials, 40 treatment comparisons and 1,804 mostly healthy adults given 30 to 1,218 mg of flavanols a day for two to 18 weeks. Systolic pressure fell by 1.76 mmHg and diastolic by 1.76 mmHg against control. That is real, statistically significant and small, and the reviewers could not find a single trial of long-term daily use.

The one large trial, and what it did not find

The best evidence on hard outcomes is COSMOS, a randomized, placebo-controlled trial of 21,442 older US adults published in 2022. Half took a daily cocoa extract supplying 500 mg of flavanols, including 80 mg of epicatechin, for a median of 3.6 years. On the primary endpoint, total cardiovascular events, the extract did not reach significance: a hazard ratio of 0.90 with a confidence interval of 0.78 to 1.02. On a prespecified secondary endpoint, death from cardiovascular causes, it did: a 27 percent reduction, hazard ratio 0.73, interval 0.54 to 0.98. Among people who actually took their pills, total events fell as well. The trial was funded by an investigator-initiated grant from Mars Edge alongside National Institutes of Health grants, and Mars supplied the capsules.

Read plainly: a purified flavanol dose roughly two and a half times the EFSA threshold, taken daily for years, produced a promising secondary result and a primary result that missed. That is the strongest cardiovascular evidence that exists, and it concerns capsules, not bars.

Why the cohort studies flatter chocolate

The numbers that make headlines are bigger, and they come from observation. A 2011 BMJ meta-analysis of seven studies and 114,009 people found the highest chocolate eaters had 37 percent less cardiovascular disease and 29 percent less stroke than the lowest, and its authors said outright that the finding needed trials to confirm it. In 2024 the same journal published an analysis of 192,208 people in three Harvard cohorts followed for about 25 years: five or more servings of dark chocolate a week went with a 21 percent lower risk of type 2 diabetes, milk chocolate went with no reduction and with long-term weight gain, and the authors wrote that the observational design "prevents firm conclusions from being drawn about causality."

People who eat dark chocolate several times a week and stay lean over 25 years tend to differ from everyone else in ways no questionnaire fully captures. The authors adjusted for diet and lifestyle and still said other factors could explain the result. That is the problem with every cohort finding on chocolate and the reason the trial results are smaller.

The brain

Here the evidence is unusually clean and unusually negative. The cognitive arm of the same COSMOS trial gave older adults 500 mg of cocoa flavanols a day for three years and found no benefit to global cognition, episodic memory or executive function. The multivitamin arm of the same study did show an effect; the cocoa arm did not. Claims that dark chocolate sharpens memory or mood rest on short trials and small samples, and the one long trial says no.

The evidence, graded

Claim Best evidence Kind Grade What it means for a bar
Helps blood vessels dilate EFSA opinion, 2012 Randomized trials, reviewed Strong Needs 200 mg flavanols a day; ordinary bars do not state their content
Lowers blood pressure Cochrane, 2017 35 randomized trials Strong but small About 2 mmHg over weeks; no long-term trials
Fewer heart attacks and strokes COSMOS, 2022 One randomized trial, 21,442 people Unproven Primary endpoint missed; capsules, not chocolate
Fewer cardiovascular deaths COSMOS, 2022 Secondary endpoint Promising 27 percent reduction; needs replication
Lower diabetes risk BMJ cohorts, 2024 Observational Association only Dark yes, milk no; cannot show cause
Sharper memory or mood COSMOS-Mind, 2023 Randomized trial, 3 years No effect found Enjoy it for other reasons
Contains lead and cadmium Consumer Reports 2022; Frontiers in Nutrition 2024 Product testing Strong Real; varies by origin and maker; see below

The bar is not the supplement

Three things stand between the trial dose and the bar in your hand. First, no wrapper lists flavanol content, and the cacao percentage does not predict it. The percentage tells you how much of the bar came from the bean, not what survived roasting, fermentation and conching, as the piece on what the percentage means explains. Second, processing destroys flavanols. A 2008 study of commercial cocoa powders found natural cocoa averaged 34.6 mg of flavanols per gram, lightly alkalized 13.8, medium 7.8 and heavily alkalized 3.9, a linear fall as pH rose. Any bar or cocoa labeled "processed with alkali" or "Dutched" has given most of them up. Third, the trial products were engineered to keep flavanols. EFSA's 10 g figure is for high-flavanol chocolate made for the purpose, not a standard bar.

Then there is the arithmetic. USDA data for chocolate at 70 to 85 percent cacao give 598 calories, 42.6 g of fat, 24.5 g of it saturated, 24 g of sugar and 10.9 g of fiber per 100 g, with 11.9 mg of iron, 228 mg of magnesium, 80 mg of caffeine and 802 mg of theobromine. Scaled to a one-ounce square, about 30 g, that is roughly 180 calories, 13 g of fat, 7 g of sugar, 3 g of fiber, 3.6 mg of iron, 68 mg of magnesium, 24 mg of caffeine and 240 mg of theobromine. The largest share of the saturated fat is stearic acid, which a 2010 systematic review found does not raise LDL cholesterol the way other saturated fats do. The iron and magnesium are genuine. So are the calories: an ounce a day adds roughly 1,260 calories to the week.

The metals

In 2022 Consumer Reports tested 28 dark chocolate bars and found lead and cadmium in all of them; 23 exceeded its daily threshold for at least one metal at an ounce a day, and five exceeded both. The thresholds were California's Proposition 65 maximum allowable dose levels, 0.5 micrograms of lead and 4.1 micrograms of cadmium a day, which are among the strictest standards in use. A 2024 study in Frontiers in Nutrition tested 72 cocoa products bought over eight years and found 43 percent over the lead threshold and 35 percent over the cadmium threshold, with organic-labeled products no better, and with levels falling in later years. Cadmium comes up from the soil into the bean, so it tracks origin; FDA testing has found Latin American cacao runs higher than African. Lead lands on beans after harvest, as dust during sun-drying, so it tracks handling. Neither is a reason to fear a bar. Both are a reason not to treat dark chocolate as a daily supplement, and a reason to ask a maker where the beans came from. The buying piece covers what a label can and cannot tell you.

Short answers to the rest

Is 85 percent healthier than 70? Less sugar per ounce, more fat, and no wrapper will tell you the difference in flavanols. Does raw cacao have more? Unfermented and unroasted beans are generally reported to keep more flavanols, but "raw" on a label is not a regulated term; the cocoa or cacao piece covers what the word does and does not promise. Does milk chocolate count? Only in the cohort study, where it went with no lower diabetes risk and with weight gain. Does adding milk block the flavanols? One 2003 study said so and later trials did not confirm it; the question is open. How much would match the trials? Nobody can say from a wrapper, which is the honest answer to the whole question. Should children eat dark chocolate daily? The metals studies are the reason to keep it occasional for small bodies.

The single-origin bars on our shelf, from Dandelion's Costa Esmeraldas to Amano's Ocumare, are sold as food, not as medicine, at 263 Josephine Street in Cherry Creek, Denver, and shipped nationwide from theambrosianpantry.com. Visitors can ask staff questions in the shop. The right dose is a square that tastes of somewhere.

Sources & Further Reading

Back to blog