Sleep deprivation is increasingly recognized as a public health challenge. While several studies have provided evidence of important associations between sleep deprivation and health outcomes, it is less clear whether sleep deprivation is a cause or a marker of poor health. This paper studies the causal effects of sleep on health status and obesity exploiting the relationship between sunset time and circadian rhythms and the discontinuities in sunset time created by time zone boundaries. Using data from the American Time Use Survey, we show that individuals living in counties on the eastern side of a time zone boundary go to bed later and sleep less than individuals on the opposite side of the time zone boundary. These findings are driven by individuals whose biological schedules and waking up times are constrained by social schedules (i.e., work schedules, school starting times) which respond to returns to coordination and are not affected by solar cues. Indeed, we find that discontinuities in sleeping time are largest among people working in the public administration, health and school sectors, and lowest in the retail and wholesale industry. Exploiting these discontinuities, we find evidence that sleep deprivation increases the likelihood of reporting poor health status and the incidence of obesity. Our results suggest that the increase in obesity is explained by both changes in eating behavior and a decrease in physical activity.