This study examines the relationship between long-term intake of six flavonoid classes and incidence of CVD and CHD, using a comprehensive flavonoid database and repeated measures of intake, while accounting for possible confounding by components of a healthy dietary pattern. Flavonoid intakes were assessed using a FFQ among the Framingham Offspring Cohort at baseline and three times during follow-up. Cox proportional hazards regression was used to characterise prospective associations between the natural logarithms of flavonoid intakes and CVD incidence using a time-dependent approach, in which intake data were updated at each examination to represent average intakes from previous examinations. Mean baseline age was 54 years, and 45 % of the population was male. Over an average 14·9 years of follow-up among 2880 participants, there were 518 CVD events and 261 CHD events. After multivariable adjustment, only flavonol intake was significantly associated with lower risk of CVD incidence (hazard ratios (HR) per 2·5-fold flavonol increase=0·86, P
trend=0·05). Additional adjustment for total fruit and vegetable intake and overall diet quality attenuated this observation (HR=0·89, P
trend=0·20 and HR=0·92, P
trend=0·33, respectively). There were no significant associations between flavonoids and CHD incidence after multivariable adjustment. Our findings suggest that the observed association between flavonol intake and CVD risk may be a consequence of better overall diet. However, the strength of this non-significant association was also consistent with relative risks observed in previous meta-analyses, and therefore a modest benefit of flavonol intake on CVD risk cannot be ruled out.