Residency programs have been required to limit duty hours for housestaff since 2003. The effect of this policy change on errors remains controversial, although at least one prior study demonstrated fewer errors when housestaff worked shorter shifts. In this survey of internal medicine residents, the number of hours worked was not associated with a perceived risk of committing medical errors. Administrative workload and overall work stress were more closely associated with suboptimal care. The results suggest that, in order to reduce errors committed by residents, both the number of hours worked and the nature of residents' work should be addressed.