The efficacy of scheduled microbreaks in controlling fatigue in a highly repetitive computer task was evaluated. Twenty experienced female data entry operators (mean age, 27.4 years) participated in the study. On each of 2 days, subjects worked at a data entry task for three 40 minute periods each morning and afternoon. After each 20 minutes of work, a self regulated microbreak was given, lasting until the subject felt ready to continue work. Microbreak length was measured by the computer, and subjects' mood was monitored by a survey administered at the beginning of each 40 minute period. Task performance was measured as keystroke output, error rate, and self correction rate, and subject heart rate and interbeat interval were continuously monitored. Mean microbreak duration was 27.4 seconds. Keystroke output was reduced and correction rate increased, both significantly, between the first and second halves of each work period. No change in error rate or the cardiac parameters was observed. Long microbreak duration predicted low correction rates and longer mean interbeat intervals in the second half of the work period. High first half correction rates were predictive of long microbreak length. Long microbreaks were associated with the mood indicators fatigue and boredom. Within each three period session, significant variation was found in microbreak duration, cardiac response, and mood indicators. The authors conclude that microbreak duration is positively correlated with worker perception of fatigue, although the efficacy of scheduled microbreaks of discretionary duration may be limited in combating fatigue.