Recent epidemiologic studies have found that self-reported duration of sleep is associated with obesity, diabetes, hypertension, and mortality. The extent to which self reports of sleep duration are similar to objective measures and whether individual characteristics influence the degree of similarity are not known. Eligible participants at the Chicago site of the Coronary Artery Risk Development in Young Adults Study were invited to participate in a 2003-2005 ancillary sleep study; 82% (n = 669) agreed. Sleep measurements collected in 2 waves included 3 days each of wrist actigraphy, a sleep log, and questions about usual sleep duration. We estimate the average difference and correlation between subjectively and objectively measured sleep by using errors-in-variables regression models. Average measured sleep was 6 hours, whereas the average from subjective reports was 6.8 hours. Subjective reports increased on average by 34 minutes for each additional hour of measured sleep. Overall, the correlation between reported and measured sleep duration was 0.47. Our model suggests that persons sleeping 5 hours over-reported their sleep duration by 1.2 hours, and those sleeping 7 hours over-reported by 0.4 hours. The correlations and average differences between self-reports and measured sleep varied by health, sociodemographic, and sleep characteristics. In a population-based sample of middle-aged adults, subjective reports of habitual sleep are moderately correlated with actigraph-measured sleep, but are biased by systematic over-reporting. The true associations between sleep duration and health may differ from previously reported associations between self-reported sleep and health.