Abstract: Objective To explore the relationship between sleep duration and fasting glucose,and to provide evidence for perfecting primary prevention measures on diabetes mellitus.Methods With multistage stratified cluster random sampling,we selected some local residents aged 18 to 75 years in Xuzhou city and attained their data about sleep duration,general characteristics,behavior factors,disease conditions and physical examinations.Then we analyzed the relationship between sleep time and fasting glucose wih logistic regressions.Results Of the 15 145 residents included in the final analysis,7 557 were males,7 588 females,4 178 from urban areas,and 10 967 from rural areas.The rates of impaired fasting glucose (IFG) in the groups whose sleep duration were <4,4-5,6-7 and ≥8 hours per night were 15.3%,6.6%,3.3%,and 6.2%,respectively,with statistically siganificant difference among the groups(P<0.05).Logistic regression results showed that,after adjusting for age,hypertention,body mass index,family history of diabetes,physical activities,smoking,and alcohol drinking,compared with sleep time of 6-7 hours per night,the odds ratio (OR) and 95% confidence interval of sleep time<4,4-5 and ≥8 hours per night for IFG were 2.756(1.710-4.443),1.234(0.978-1.558),and 1.887(1.505-2.366),respectively.Conclusion Both short sleep duration and long sleep duration were risk factors for impaired fasting glucose.