To investigate the role of longitudinal change of sleep patterns in the incidence of cardiovascular diseases (CVD). Based on UK Biobank, a total of 18 172 participants were enrolled. Five dimensions of healthy sleep including early chronotype, sleep 7-8 hours/day, free of insomnia, no snoring, and no frequent excessive daytime sleepiness were used to generate a healthy sleep score (HSS) ranging from 0 to 5. Corresponding to the HSS of 0-1, 2-3, and 4-5, the poor, intermediate, and healthy sleep patterns were defined. Based on changes in HSS across assessments 1 and 2, we calculated the absolute difference of HSS. For the change in sleep patterns, we categorized five profiles (stable healthy, worsening, stable intermediate, optimizing, and stable poor sleep patterns). The outcomes were incidence of CVD including coronary heart disease (CHD) and stroke. We assessed the adjusted hazard ratios and 95% confidence intervals (CIs) by Cox hazard models. Compared with participants with stable poor patterns, those who improved their sleep patterns or maintained healthy sleep patterns had a 26% and 32% lower risk of CVD, respectively. Stable healthy sleep pattern was associated with a 29% and 44% reduced risk of CHD and stroke. Per unit, longitudinal increment of the HSS was related to an 8% lower risk of CVD and CHD. Compared with individuals with constant HSS, those with decreased HSS had a 13% higher risk of developing CVD. Optimizing sleep patterns and maintaining a healthy sleep pattern may reduce the risk of CVD.