Abstract: Objective To investigate the association of sleep duration with hypertension by meta-analysis,aiming to justify primary prevention of hypertension. Methods From PubMed,Cochrane Library,CNKI,VIP and Wanfang databases since their establishment until February 2012,a cross-sectional search for the link between sleep duration and hypertension was performed and supplemented by other retrieval methods. Then the related data were selected,evaluated,and extracted by two investigators independently according to Cochrane system. Publication bias was evaluated,and sensitivity and meta-analyses were conducted with Stata11.0. Results A total of 24studies were collected where 264 978adults were included. The Meta-analysis showed that short sleep duration was associated with a higher risk of hypertension（sleep duration≤5h:OR1.19,95% CI 1.04-1.36;≤6h:OR1.28,95% CI 1.11-1.47,both P＜0.05）,while long sleep duration also increased the risk of hypertension（sleep duration≥9h:OR1.13;95%CII 1.04-1.23,P＜0.05）. No evidence suggested any relation between higher risk of hypertension and the sleep duration≤7h（OR1.12,95%CI0.88-1.42）. Heterogeneity analysis indicated that the geographical location was one of the heterogeneity resources. No publication bias was found by Begg’s and Egger’s tests. These results of meta-analysis were of better stability by sensitivity analysis. Conclusion A U-shaped relationship between sleep duration and hypertension was found among adults through cross-sectionally observational study.