Objective: To examine the relationship between sleep duration and mortality and to quantify the likely impact of residual confounding due to poor health status on any observed association. Methods: The sample included 227,815 Australian adults aged 45 years and older recruited from 2006– 2009 (the 45 and Up Study). Sleep duration and relevant covariates (e.g., health status, demographic factors) were assessed through a self-report questionnaire. These data were linked with mortality data from the New South Wales Registry of Births, Deaths, and Marriages up to December 2010 (mean follow-up period, 2.8 y). Cox proportional hazards models examined the relationship between sleep duration and all-cause mortality adjusting for relevant sociodemographic covariates (e.g., age, gender, marital status), with further stratification by baseline health status based on physical functioning and preexisting disease. Results: The adjusted mortality risk was significantly higher in individuals reporting(hazard ratio [HR], 1.13[1.01–1.25]) and ≥10 hours of sleep (HR, 1.26[1.16–1.36]), compared to those reporting 7 hours of sleep per night. These associations differed by baseline health status (p[interaction] = 0.026) such that there was no significant relationship of sleep duration to mortality in those with good health at baseline. Conclusion: Following careful prospective controlling for baseline health, mortality risk does not significantly vary according to sleep duration. Previous findings suggesting a relationship between sleep duration and mortality could be affected by residual confounding by poor preexisting health, as reflected by a combination of preexisting illnesses and functional limitations.