Although regulations on work hours usually include a minimum weekly rest period, there is little empirical evidence concerning recovery from work. Shift-working nurses (N = 61) used a handheld computer for 28 days to complete self-ratings, cognitive-performance tasks, and a sleep diary. Many measures were worse on rest days that followed a night shift rather than a day shift and tended to be worse on first rest days compared with subsequent rest days. Alertness was lowest on the 1st rest day following a night shift. Social satisfaction was better on workdays that were preceded by 2 rather than 1 rest day. Reaction time decreased over consecutive night shifts and tended to increase over rest days following night shifts. The results are interpreted as being consistent with the combined adaptive costs of fatigue and adjustment to and from a nocturnal routine. The practical implications for scheduling rest days are considered.