BackgroundConsumption of more plant-based foods is gaining popularity, but the role of healthy versus unhealthy plant-based diets in cardiovascular disease (CVD) risk remains inconclusive. ObjectivesWe investigated associations of plant-based diet indices (PDIs) with incident CVDs in a prospective cohort study and conducted an updated meta-analysis. MethodsWe included 3507 men and 5345 women of the population-based Rotterdam Study. Multivariable Cox proportional hazard models were used to estimate hazard ratios (HRs) and 95 % confidence intervals (CIs) for CVD, coronary heart disease (CHD) and stroke with per SD increment of an overall PDI, healthy PDI (hPDI) and unhealthy PDI (uPDI), among men and women separately. We combined our findings with previously published effect estimates in an updated meta-analysis. ResultsWe documented 2015 CVD cases (1231 CHD and 952 stroke) during 107,290 person-years follow-up. Among men, the PDI and hPDI were associated with a 7 % (HR 0.93, 95 % CI 0.87–0.99) and 8 % (HR 0.92, 95 % CI 0.86–0.98) lower CVD risk. Among women, there was evidence suggesting a U-shaped association of the PDI with stroke (pnon-linearity < 0.01). In meta-analyses including up to 43,067 incident CVD cases among 359,740 participants from nine studies, the PDI and hPDI, were associated with a lower CVD risk, while uPDI with a higher CVD risk (pooled HRs [95 % CI], per SD, PDI: 0.94 [0.91–0.97], I2 = 50.4 %; hPDI: 0.94 [0.91–0.98], I2 = 74.7 %; uPDI: 1.03 [1.01–1.06], I2 = 49.0 %). ConclusionsOur findings support recommendations to consume relatively more healthy plant-based foods for CVD prevention. Potential differences by sex and non-linear associations warrant further investigation.