Multiple randomised controlled trials have shown high doses of cocoa to reduce blood pressure and improve endothelial function. However, evidence regarding long-term consumption of cocoa and its potential effect on hypertension is lacking. We aimed to prospectively evaluate if cocoa intake from various food sources was associated with incident hypertension. Among 45,653 women of the E3N cohort, chocolate consumption was estimated from a 208 item dietary questionnaire and 24-h recall. Quantities of cocoa for certain foods including chocolate drinks, Danish pastries, chocolate biscuits, chocolate cakes, chocolate candy-bars, plain chocolate bars, and chocolate desserts, were estimated using a detailed food composition table. Using Cox models with time-update exposures, we assessed associations between specific sources of cocoa, and hypertension risk. Self-reported cases were validated using a drug reimbursement database. 12,793 cases of hypertension were identified. Median cocoa consumption in the entire cohort was 2.3 g/day at baseline. Moderate but not high cocoa consumption from all sources was inversely associated with the risk of hypertension (hazard ratios HRQ1–Q3 0.93 [0.88:0.98], HRQ1–Q4 0.98 [0.93:1.03], p for trend < 0.01). Consumption of cocoa from plain chocolate was associated with reduced risk (HRT1–T2 0.89 [0.85:0.94] and HRT1–T3: 0.93 [0.89:0.97], p for trend < 0.01). Cocoa from dessert sources was associated with an increased risk of hypertension (HRT1–T3 1.09 [1.04:1.14], p for trend < 0.01). Moderate consumption of cocoa from plain chocolate could potentially reduce hypertension risk. Cocoa from sweet sources was associated with increased risk of hypertension.