A dose-response meta-analysis was conducted to summarize evidence from prospective cohort studies about the association of nighttime sleep duration and 24-hour sleep duration with risk of all-cause mortality among adults. Pertinent studies were identified by a search of Embase and PubMed databases to March 2015. A two-stage random-effects dose-response meta-analysis was used to combine study-specific relative risks and 95% confidence intervals [RRs (95% CIs)]. Thirty-five articles were included. Compared with 7 hours/day, the RRs (95% CIs) of all-cause mortality were 1.07 (1.03-1.13), 1.04 (1.01-1.07), 1.01 (1.00-1.02), 1.07 (1.06-1.09), 1.21 (1.18-1.24), 1.37 (1.32-1.42) and 1.55 (1.47-1.63) for 4, 5, 6, 8, 9, 10 and 11 hours/day of nighttime sleep, respectively (146,830 death cases among 1,526,609 participants), and the risks were 1.09 (1.04-1.14), 1.05 (1.02-1.09), 1.02 (1.00-1.03), 1.08 (1.05-1.10), 1.27 (1.20-1.36), 1.53 (1.38-1.70) and 1.84 (1.59-2.13) for 4, 5, 6, 8, 9, 10 and 11 hours/day of 24-hour sleep, respectively (101,641 death cases among 903,727 participants). The above relationships were also found in subjects without cardiovascular diseases and cancer at baseline, and other covariates did not influence the relationships substantially. The results suggested that 7 hours/day of sleep duration should be recommended to prevent premature death among adults.