The stress and long working hours of medical residency have become the basis for controversy over whether current training structures and processes adversely affect residents' skills and well-being and the quality of patient care. The authors measured cognitive and skills performances of 45 sleep-deprived pediatrics residents by using questions like those on the pediatrics board certification examination and using patient-care tasks that required coordination and dexterity. The residents were randomly divided into two groups--one stayed awake for 24 hours, the other for 34 hours--and were tested on cognitive and skills performances before and after sleep deprivation. Sleep deprivation did not have a significant effect on cognitive performance. Of the three skills tested, the residents overall needed more time to perform umbilical artery catheterization, but the group deprived of sleep for 34 hours performed vein cannulation more quickly than the group deprived for 24 hours. Implications for these findings are discussed in the context of the ongoing controversy over the structure and process of medical education.