This study assessed the effects that elevating body temperature had on sleep structure in the third and fourth sleep cycles, cycles typically characterized by a high propensity for REM sleep and diminished levels of delta amplitude and incidence. The sleep of eight women and two men was interrupted for 30 min on each of 3 consecutive nights following an undisturbed adaptation night. The subjects were awakened each night following the end of the second REM sleep period. On 2 nights, subjects were immersed to midthorax in water at either 34 degrees C (TW condition) or 41 degrees C (HW condition) for 20 min. A third interruption without immersion (NW condition) was performed to provide a second type of baseline condition. The HW condition induced a mean tympanic temperature rise of 2.5 degrees C, that returned to baseline levels in approximately 60 min. Analysis of sleep patterns focused on the two sleep cycles following interruption. The mean of the two baseline conditions (TW + NW/2) was compared with the HW condition. Sleep onset latency, REM latency, REM duration, and eye movement activity in REM were unaffected by heating. Heating evoked increases in both total NREM and slow wave sleep, though these increases were delayed until the second cycle following sleep onset (i.e., appearing in the fourth, but not the third, NREM period). These were paralleled by increases in two objective measures of delta activity: integrated slow-wave amplitude (33% increase) and slow-wave density (10% increase).