This paper investigates the relationship between health outcomes and variations in staffing levels as approximated by admissions on weekdays versus admissions on weekends. Because days of admission are potentially endogenous, we instrument on emergency admissions only, which are reasonably exogenous to the time of admission. Further, we introduce a direct measure for within-diagnosis variation in severity across days of admission to control for the unobservable selection of patients. We find that after controlling for patient heterogeneity and endogeneity of the day of admission there is still a significant variation in mortality rates between weekend and weekday admissions. Patients admitted during the weekend exhibit higher in-hospital mortality rates. We also find signs of premature discharge, as patients with short lengths of stay tend to exhibit higher probability to be readmitted as emergency cases.