In patients with obstructive sleep apnea, it is believed that body position influences apnea frequency. Sleeping in the lateral decubitus position often results in significantly fewer apneas, and some have recommended sleeping on the side as the major treatment intervention. Previous studies, although calculating apnea-hypopnea index (AHI) for supine and lateral decubitus positions, have not taken sleep stage into account. To examine the effect of both sleep stage and body position on apnea duration (AD) and frequency, we determined AHI and AD in all spontaneous body positions during rapid eye movement (REM) and non-REM (NREM) sleep by reviewing videotapes and polysomnograms from 11 overnight studies of 7 obese patients with severe sleep apnea. Consistent with previous work, AD was significantly longer in REM then in NREM (32.5 +/- 2.3 s versus 23.5 +/- 1.9 s; p less than 0.05). This difference persisted when adjusting for body position. AHI was greater on the back than on the sides (84.4 +/- 4.9/h versus 73.6 +/- 7.5/h, p less than 0.05), but after accounting for sleep stage, this difference remained only for NREM (103 +/- 4.8/h versus 80.3 +/- 9.2/h, p less than 0.05) and not for REM (83.6 +/- 5.3/h versus 71.1 +/- 4.2/h, p NS). Although reduced, AHI on the sides still remained clinically very high. Body position changed frequently throughout the night, but some patients spent little or no time on their back. We conclude that AD is longer in REM than NREM, regardless of position, and AHI is higher on the back only in NREM. As AHI remains very high on the sides, favoring the lateral decubitus position may not be as beneficial as previously thought in very obese patients. Less obese patients are more likely to benefit by position changes.