Background and objectives: The association between sleep duration and obesity in children and adolescents has been widely evaluated, whereas the current findings are mixed and prospective studies are limited. To shed more light on this issue and explore the dose-response relationship, we performed the present updated meta-analysis by synthesizing the results of prospective cohorts. Methods: Literature retrieval, study selection and data extraction were completed independently and in duplicate. Effect-size estimates are expressed as relative risk (RR) with 95% confidence interval (CI) or standardized regression coefficient (beta) with standard error. Results: Data from 33 articles, involving 57,848 children and adolescents, were meta-analyzed. Overall analyses revealed statistically significant associations of short (adjusted RR = 1.57, 95% CI: 1.36 to 1.81, P < 0.001) and long sleep duration (0.83, 0.75 to 0.93, 0.001) with obesity. Short sleep duration was also associated with significant changes in body mass index z-score (mean difference = -0.06; 95% CI: -0.09 to -0.04; P < 0.001). By contrast, long sleep duration was identified as a protective factor for childhood obesity. In dose-response analyses, short sleep duration was significantly associated with obesity in toddlers (1-2 years) (adjusted RR = 1.20, 95% CI: 1.07 to 1.34, P = 0.001), preschool-aged (3-5 years) children (1.58, 1.36 to 1.83, <0.001), and school-aged (6-13 years) children (1.82, 1.51 to 2.21, <0.001). In subgroup analyses, geographic region, sleep duration assessment, age, and follow-up interval were possible sources of heterogeneity. Conclusions: Our findings indicate that short sleep duration can increase the risk of obesity in children and adolescents, especially within 3-13 years of age, and long sleep duration seemed beneficial in preventing obesity. (C) 2020 Elsevier B.V. All rights reserved.