Although polycythemia of high altitude is usually due to excessive hypoxemia, in some patients the hematocrit is elevated out of proportion to the degree of hypoxemia measured awake. One possible explanation is that severe hypoxemia occurs during sleep in these subjects. We therefore monitored oxygen saturation (SaO2), breathing pattern, and electroencephalogram (EEG) during sleep in five normal high-altitude residents and in five patients with excessive polycythemia. The polycythemic patients were studied as part of a placebo--drug double-blind crossover trial of the respiratory stimulant drug medroxyprogesterone acetate (MPA). The polycythemic patients while taking placebo were much more hypoxemic during sleep than the normals (all-night mean SaO2: 79.4 +/- 1.7% versus 87.8 +/- 1.7%, p less than 0.01). Abnormalities in breathing patterns were observed in all the subjects, especially during REM stage sleep. In polycythemic subjects, this resulted in precipitous hypoxemia with SaO2 as low as 50%--70%. Severe hypoxemia was not observed in control subjects despite similar abnormalities in breathing. Significant improvement in nocturnal SaO2 occurred when the polycythemic patients were taking MPA, mean SaO2 rising from 79.4 +/- 1.7% to 83.7 +/- 0.7%, p less than 0.05. Of probably greater importance, MPA largely prevented the precipitous drops in SaO2, mean lowest SaO2 rising from 64.6 +/- 4.7% to 76.0 +/- 2.1% p less than 0.05. The severe decreases in SaO2 during sleep may explain elevations in hematocrit that are out of proportion to the awake SaO2 in man at high altitude. The therapeutic effect of MPA in this condition may be due to amelioration of sleep hypoxemia.