Objectives Continuous positive airway pressure (CPAP) improves sleep apnea survival. We tested whether CPAP is associated with better survival than uvulopalatopharyngoplasty (UPPP). Study design and methods This retrospective cohort database study included all sleep apnea patients treated with CPAP or UPPP in Veteran Affairs facilities from October 1997 through September 2001. Treatment groups were compared with Cox regression, adjusting for age, gender, race, year treatment was initiated, and comorbidity. Sleep apnea severity and CPAP use data were not available. Results By September 2002, 1339 (7.1%) of 18,754 CPAP patients and 71 (3.4%) of 2,072 UPPP patients were dead (P < 0.001). After adjustment, CPAP patients had 31% (95% confidence interval, 3% to 67%, P = 0.03) higher probability of being dead at any time, relative to UPPP patients. Conclusions UPPP confers a survival advantage over CPAP, after adjustment for age, gender, race, year of treatment, and comorbidity. However, we were unable to adjust for sleep apnea severity or CPAP use. Surgical treatment should be considered in sleep apnea patients who use CPAP inadequately.