Background:Sleep duration is dependent on circadian rhythm that controls a variety of key cellular functions. Circadian disruption has been implicated in colorectal tumorigenesis in experimental studies. We prospectively examined the association between sleep duration and risk of colorectal cancer (CRC).Methods:In the Women's Health Initiative Observational Study, 75 828 postmenopausal women reported habitual sleep duration at baseline 1993-1998. We used Cox proportional hazards regression model to estimate the hazard ratio (HR) of CRC and its associated 95% confidence interval (CI).Results:We ascertained 851 incident cases of CRC through 2010, with an average 11.3 years of follow-up. Compared with 7 h of sleep, the HRs were 1.36 (95% CI 1.06-1.74) and 1.47 (95% CI 1.10-1.96) for short (5h) and long (9h) sleep duration, respectively, after adjusting for age, ethnicity, fatigue, hormone replacement therapy (HRT), physical activity, and waist to hip ratio. The association was modified by the use of HRT (P-interaction=0.03).Conclusion:Both extreme short and long sleep durations were associated with a moderate increase in the risk of CRC in postmenopausal women. Sleep duration may be a novel, independent, and potentially modifiable risk factor for CRC.British Journal of Cancer advance online publication, 3 January 2013; doi:10.1038/bjc.2012.561 www.bjcancer.com.