Background: The prevalence of undiagnosed obstructive sleep apnea (OSA) during preoperative evaluation and the best method to screen OSA and its association with postoperative complications remain unclear. Objectives: To determine the prevalence of undiagnosed OSA in preoperative Indian patients undergoing noncardiac surgery, to compare the diagnostic accuracy of the STOP-BANG questionnaire to a preoperative level III sleep study, and to assess the association of OSA with postoperative complications. Methods: A prospective cohort of 245 consecutive adults with ≥2 risk factors for OSA who underwent noncardiac surgery between July 2011 and February 2013 were studied. The STOP-BANG questionnaire was administered to all patients, and 182/245 (74.2%) patients underwent a preoperative level III sleep study. Patients were followed for postoperative complications in hospital and contacted at 30 days after surgery. Results: 70/182 (38.5%) obtained a new diagnosis of OSA, including 11/182 (6%) with moderate to severe OSA (apnea-hypopnea index ≥15/h). On logistic regression analyses, the presence of OSA was independently associated with postoperative oxygen desaturation (OR 5.96, 95% CI 2.35-15.1, p < 0.01), a postoperative complication within 7 days (OR 3.63, 95% CI 1.77-7.45, p < 0.01) and within 30 days (OR 3.5, 95% CI 1.74-7.1, p < 0.01). The STOP-BANG questionnaire did not identify 12/70 (17%) of the patients diagnosed with OSA and classified 28% of the cohort as OSA when the level III sleep study was negative. Conclusions: Unrecognized OSA is common in preoperative patients and is independently associated with postoperative complications. The STOP-BANG questionnaire had a lower performance in the diagnosis of OSA in a South Indian population than the level III sleep study.