Abstract Background and Aims Evidence on the link between sleep patterns and cardiovascular diseases (CVDs) in the community essentially relies on studies that investigated one single sleep pattern at one point in time. This study examined the joint effect of five sleep patterns at two time points with incident CVD events. Methods By combining the data from two prospective studies, the Paris Prospective Study III (Paris, France) and the CoLaus|PsyCoLaus study (Lausanne, Switzerland), a healthy sleep score (HSS, range 0–5) combining five sleep patterns (early chronotype, sleep duration of 7–8 h/day, never/rarely insomnia, no sleep apnoea, and no excessive daytime sleepiness) was calculated at baseline and follow-up. Results The study sample included 11 347 CVD-free participants aged 53–64 years (44.6% women). During a median follow-up of 8.9 years [interquartile range (IQR): 8.0–10.0], 499 first CVD events occurred (339 coronary heart disease (CHD) and 175 stroke). In multivariate Cox analysis, the risk of CVD decreased by 18% [hazard ratio (HR) 0.82, 95% confidence interval (CI) 0.76–0.89] per one-point increment in the HSS. After a median follow-up of 6.0 years (IQR: 4.0–8.0) after the second follow-up, 262 first CVD events occurred including 194 CHD and 72 stroke. After adjusting for baseline HSS and covariates, the risk of CVD decreased by 16% (HR 0.84, 95% CI 0.73–0.97) per unit higher in the follow-up HSS over 2–5 years. Conclusions Higher HSS and HSS improvement over time are associated with a lower risk of CHD and stroke in the community.