Objective The association between high-density lipoprotein cholesterol (HDL-C) levels and mortality remains controversial. We aimed to investigate the potential dose–response associations between HDL-C levels and mortality from all causes, cardiovascular disease and cancer in the general population. Methods PubMed and Embase were searched through April 2019. Prospective cohort studies reporting risk estimates of HDL-C levels and mortality were included. Linear and non-linear dose–response analyses were conducted. A random-effects model was employed to calculate pooled hazard ratio. Results Thirty-seven studies, involving 3,524,505 participants and more than 612,027 deaths, were included. HDL-C level was found to be associated with mortality from all causes, cardiovascular disease and cancer in a J-shaped dose–response pattern, with the lowest risk observed at HDL-C levels of 54–58 mg/dL, 68–71 mg/dL and 64–68 mg/dL, respectively. Compared with HDL-C level of 56 mg/dL, the pooled hazard ratios for all-cause mortality were 1.03 (95% confidence interval (CI) 1.01, 1.05) and 1.10 (95% CI 1.09, 1.12) for each 10-mg/dL increase and decrease in HDL-C levels, respectively; furthermore, compared with the reference category, the pooled hazard ratios for all-cause mortality were 1.21 (95% CI 1.09, 1.36) and 1.36 (95% CI 1.21, 1.53) for the highest and the lowest categories of HDL-C levels, respectively. Similar results were obtained for cardiovascular and cancer mortality. Conclusions In the general population, HDL-C level is associated with mortality from all causes, cardiovascular disease and cancer in a J-shaped dose–response manner; both extremely high and low HDL-C levels are associated with an increased risk of mortality.