Guidelines for the performance of cardiopulmonary resuscitation (CPR) have been revised recently and now advocate that chest compressions are performed without interruption for 3 min in patients during asystole and pulseless electrical activity. The aim of the present study was to determine if rescuer fatigue occurs during 3 min of chest compressions and if so, the effects on the rate and quality of compressions. Forty subjects competent in basic life support (BLS) were studied. They performed continuous chest compressions on a Laerdal Skillmeter Resusci-Anne manikin for two consecutive periods of 3 min separated by 30 s. The total number of compressions attempted was well maintained at approximately 100 min(-1) throughout the period of study. However, the number of satisfactory chest compressions performed decreased progressively during resuscitation (P < 0.001) as follows: first min, 82 min(-1); second, 68 min(-1); third, 52 min(-1); fourth, 70 min(-1); fifth, 44 min(-1); sixth, 27 min(-1). We observed significant correlations between the number of satisfactory compressions performed and both height and weight of the rescuer. Female subjects achieved significantly fewer satisfactory compressions compared with males (P = 0.03). Seven subjects (five female, two male) were unable to complete the second 3-min period because of exhaustion. We conclude that rescuer fatigue adversely affects the quality of chest compressions when performed without interruption over a 3-min period and that this effect may be greater in females due to their smaller stature. Consideration should be given to rotating the rescuer performing chest compressions after 1 min intervals.