Dark chocolate and cocoa products have produced small average blood-pressure reductions in trials, but that does not make an ordinary bar a treatment. The product, comparison group, starting blood pressure and measured outcome all affect what the result means.
Newer evidence adds an important distinction: lowering a blood-pressure reading over weeks is different from preventing new hypertension over years. A 2025 cocoa-extract analysis found no significant prevention benefit overall, even though one subgroup showed a signal worth further study.
This guide compares those findings with the short-term trials and explains why cocoa percentage cannot supply a personal flavanol dose. It retains the difference between blood pressure, artery-function tests and actual illness.
Does dark chocolate lower blood pressure?
Some controlled trials support a modest effect from studied cocoa or chocolate products. The 2017 Cochrane review found an average reduction of roughly 2 mmHg, with considerable variation between studies.
A 2026 review by Sakuda and colleagues also found small reductions: about 2.7 mmHg systolic and 1.8 mmHg diastolic. Its included interventions lasted 14 days to six months, and evidence certainty was moderate for systolic pressure and low for diastolic pressure.
Those are group averages under particular study conditions, not a promised response to a square of chocolate. Different reviews select overlapping studies using different criteria, so their participant totals should not be added together.
The practical answer is therefore qualified: cocoa flavanols can affect vascular markers, while the evidence does not identify a universal chocolate serving that treats an individual’s blood pressure.
What the studies actually compared
Before comparing the size of an effect, check the study’s question. A chocolate-versus-white-chocolate experiment, a pooled marker analysis and a long-term capsule trial do not test the same thing.
| Evidence | Population and comparison | What changed | Important limit |
|---|---|---|---|
| Taubert 2007 randomized trial | 44 adults with above-optimal, untreated pressure; tested dark chocolate versus matching white chocolate for 18 weeks. | A small blood-pressure reduction in the dark-chocolate group. | One small sample and a particular product, not every retail bar. |
| Ried 2017 Cochrane review | 35 trials with 40 comparisons; products and controls varied. | About 2 mmHg lower blood pressure on average over two to 18 weeks. | Short-term average with heterogeneous studies. |
| Sakuda 2026 meta-analysis | 25 reports, 1,167 participants for systolic pressure; varied cocoa-containing foods and formulations. | Modest pressure reductions and some vascular-marker improvements. | No clear dose threshold; it did not establish prevention of new hypertension. |
| Hamaya 2025 COSMOS analysis | 8,905 older adults without hypertension under the study definition; cocoa extract versus placebo. | No significant overall reduction in self-reported new hypertension over a median 3.4 years. | Not a chocolate-bar trial or a treatment trial in people with established hypertension. |
The original 2007 trial used 6.3 grams of a specified dark chocolate containing 30 milligrams of polyphenols. That historical product amount is not a prescription or a reliable conversion for another bar.
Polyphenols are a broader category than flavanols. Keep the source’s actual chemical measure rather than treating every reported milligram value as interchangeable.
What the longer COSMOS analysis adds
The 2025 analysis tested standardized cocoa extract containing 500 milligrams of flavanols daily. Overall, the hazard ratio for new hypertension was 0.96, with a 95 percent confidence interval of 0.88 to 1.05.
Because that interval includes 1, the overall result does not establish a prevention benefit. It is also not a change of 0.96 mmHg or a guaranteed 4 percent reduction for a person.
A lower-baseline-systolic-pressure subgroup showed a benefit signal, but the paper describes that subgroup as post hoc and recommends caution. It should not replace the overall finding or become a rule for choosing supplements from a home reading.
Hypertension outcomes relied on self-reports rather than direct research blood-pressure measurements or record-confirmed diagnoses. The study’s definition is not a home diagnostic test, and the population was predominantly White older US adults.
The useful update is about the question answered: a small short-term shift in pressure does not automatically establish fewer new diagnoses years later. The two findings can coexist without one cancelling the other.
Blood pressure, FMD and hazard ratios use different scales
The earlier 2019 endothelial-function review reported a 1.17-percentage-point improvement in flow-mediated dilation, or FMD. FMD is an ultrasound-based artery-response measurement, not the pressure measured by a blood-pressure cuff.
The newer 2026 synthesis reported a similar direction for FMD, but rated that evidence low certainty. Neither marker result translates directly into a percentage reduction in someone’s heart-attack risk.
| Number | What the unit describes | What it does not mean |
|---|---|---|
| A few mmHg | Difference in a blood-pressure measurement. | The same percentage change in circulation or disease risk. |
| FMD percentage points | Change in the measured artery-dilation percentage. | That percentage more skin blood flow or fewer heart attacks. |
| A hazard ratio | A time-to-event comparison between groups, interpreted with its uncertainty. | A pressure reading, an absolute percentage-point benefit or a personal forecast. |

Why 70 percent cocoa cannot tell you a trial dose
Cocoa percentage describes cocoa-derived ingredients by weight, not a tested flavanol amount. Processing and product composition can change the compounds that remain in the finished food.
The older and newer studies used different beverages, powders, capsules and chocolate preparations. A label that supplies only cocoa percentage does not let you match those interventions by counting squares.
A commercial-product analysis also found that calculated cacao percentage correlated less closely with several flavanol measures than nonfat cocoa solids did. That relationship across sampled products does not reveal the measured dose in an individual untested bar.
| Information shown | What it tells you | What remains unknown |
|---|---|---|
| Cocoa percentage | The proportion of cocoa-derived ingredients. | Flavanol milligrams in the food you eat. |
| Serving weight and nutrition panel | Food amount and declared nutrients for that serving. | Equivalence to a tested cocoa formulation. |
| A measured flavanol declaration | A more specific composition claim to check against its testing and serving basis. | A guaranteed effect or a reason to copy a trial dose. |
This distinction is food-label literacy, not a product ranking. Color, bitterness and cocoa percentage should not be treated as substitutes for a chemical assay.
More is not an established treatment strategy
The 2026 review did not identify a clear minimum effective dose or optimal threshold. Its dose-response analysis could not estimate a nonlinear curve because most trials tested only one active dose.
The 2019 review modeled an apparent optimum for its particular FMD data, with substantial variation and risk-of-bias concerns. That modeling result is not an individualized amount to pursue, and a different outcome may have a different relationship.
Study certainty also matters: the 2026 review did not systematically quantify safety outcomes, and adverse-event reporting varied. A favorable average marker change does not establish that unlimited intake or a supplement is appropriate for everyone.
Funding should remain visible alongside these limitations. The 2026 review was funded by Meiji and its authors reported commercial employment; COSMOS reported NIH and industry support, with companies stating no role in its design, analysis or manuscript preparation.
What the FDA’s qualified claim covers
The FDA’s 2023 announcement concerns specified high-flavanol cocoa powder and foods containing it, with carefully qualified cardiovascular-risk wording. The agency described the supporting evidence as very limited.
That decision does not apply to regular cocoa powder or ordinary chocolate. It is not approval of a chocolate treatment or permission to turn a package’s cocoa percentage into a clinical promise.
The 70 percent skin-blood-flow claim remains unverified
A cocoa-percentage label and a relative change in skin blood flow are different measurements. The viral claim needs its exact source, product, population, duration and testing conditions before it can be interpreted.
FMD, blood pressure and local skin circulation cannot fill in for that missing study. The trials discussed here do not justify promising 70 percent more circulation from a square of chocolate.
Enjoy cocoa as food and keep care decisions separate
If you enjoy cocoa or chocolate, you can consider it within your usual food preferences rather than adding it to chase a medicinal effect. The evidence does not require buying a supplement, forcing a disliked food or aiming for a particular cocoa percentage.
Blood-pressure care should follow an appropriate assessment and your agreed treatment plan. Do not stop or adjust medication because of a chocolate study or interpret one home reading as proof that cocoa worked.
Tell your health professional about supplements and concerns about diet or symptoms. Food allergies and individual tolerances also matter when deciding what fits your circumstances.
For another example of separating intermediate markers from longer-term outcomes, see our sauna and heart-health evidence review. The same reading principle applies: keep the actual intervention and endpoint attached to the claim.
Get the next evidence check.
Five practical fitness stories worth your time, every Friday.
One email a week. No product pitches. Unsubscribe in one click.
Sources and further reading
- Taubert et al. (2007), Effects of low habitual cocoa intake on blood pressure and bioactive nitric oxide. Original small randomized trial.
- Ried, Fakler and Stocks (2017), Effect of cocoa on blood pressure. Cochrane review.
- Sun et al. (2019), Dose-response relationship between cocoa flavanols and human endothelial function. FMD review.
- Miller et al. (2009), Survey of commercially available chocolate- and cocoa-containing products. Product composition and percentage-cacao comparison.
- Hamaya et al. (2025), Long-term effect of cocoa extract supplementation on incident hypertension. COSMOS analysis.
- Sakuda et al. (2026), Impact of cocoa flavanols on blood pressure and vascular function. Updated meta-analysis published online in September 2026.
- FDA (2023), Qualified health claim for high-flavanol cocoa powder. Scope and evidence qualification.
Medical disclaimer: This article provides general nutrition and health information, not medical advice or a treatment for high blood pressure. Discuss symptoms, supplements and treatment decisions with a qualified health professional who can assess your circumstances.






