BACKGROUND: Uncertainty remains about the optimum step count per day for health promotion. OBJECTIVE: We aimed to investigate the association between step count per day and all-cause mortality risk. METHODS: PubMed, Scopus, and ISI Web of Science were searched to January 2021 to find prospective cohort studies of the association between device-based step count per day and all-cause mortality risk in the general population. Two reviewers extracted data in duplicate and rated the certainty of evidence using the GRADE approach. Study-specific hazard ratios (HRs) were pooled using a random-effects model. RESULTS: Seven prospective cohort studies with 175,370 person-years and 2310 cases of all-cause mortality were included. The HR for each 1000 steps per day was 0.88 (95% CI 0.83-0.93; I2 = 79%, n = 7) in the overall analysis, 0.87 (95% CI 0.78-0.97; I2 = 59%, n = 3) in adults older than 70 years, and 0.92 (95% CI 0.89-0.95; I2 = 37%, n = 2) in studies controlled for step intensity. Dose-response meta-analysis indicated a strong inverse association, wherein the risk decreased linearly from 2700 to17,000 steps per day. The HR for 10,000 steps per day was 0.44 (95% CI 0.31-0.63). The certainty of evidence was rated strong due to upgrades for large effect size and dose-response gradient. CONCLUSIONS: Even a modest increase in steps per day may be associated with a lower risk of death. These results can be used to develop simple, efficient and easy-to-understand public health messages.