Although obstructive sleep apnea (OSA) has been studied in detail for over a decade, the mortality of this disorder is unclear. We calculated cumulative survival in 385 male OSA patients. We found that those with an apnea index (AI) greater than 20 had a much greater mortality than those with AI = less than 20. The probability of cumulative eight-year survival was .96 +/- 0.02 (SE) for AI = less than 20 vs. 63 +/- 0.17 for AI greater than 20 (p less than .05). This difference in mortality related to AI was particularly true in the patients less than 50 years of age in whom mortality from other causes is not common. None of the patients treated with tracheostomy or nasal CPAP died. Eight of the patients treated with uvulopalatopharyngoplasty (UPPP) died and the cumulative survival of the UPPP-alone treated group was not different from the survival curve of untreated OSA patients with an apnea index of greater than 20. We conclude that OSA patients with an apnea index of greater than 20 have a greater mortality than those below 20 and that UPPP patients be restudied after therapy. If the latter patients are found not to have marked amelioration of their AI, then they should be treated by nasal CPAP or tracheostomy.