OBJECTIVE To identify pre-operative risk factors predicting complications following adenotonsillectomy in children with severe OSA. METHODS Retrospective chart review in an academic tertiary care center. Children with symptoms of OSA with overnight polysomnography (PSG) revealing apnea-hypopnea index (AHI) >10, who underwent adenotonsillectomy with overnight postoperative observation between 2008 and 2012. Univariate logistic regression was used to assess odds ratio (OR) of individual risk factors versus postoperative complications such as overnight desaturations <90%, length of stay (LOS)>24 h, supplemental oxygen requirement, and transfer to a higher level of care. RESULTS All patients (n=157) with severe OSA were observed overnight. Mean age was 5.3±3.7 years. Twenty-five (15.9%) patients had LOS>24 h. Forty-two (26.8%) had overnight desaturations <90%. AHI ≥15 and O2 saturation nadir <80% on preop polysomnography (PSG) were independent predictors of post-op O2 saturation <90% and LOS>24 h. (p<0.05). PSG minimum saturation <80% was the strongest predictor of all variables examined with an OR of 6.98 (3.15-15.48, 95% CI) for desaturation <90% and 5.19 (2.11-12.75, 95% CI) for LOS>24 h. Preop PSG O2 saturation<90% predicted overnight post op oxygen requirement with an OR of 3.38 (1.39-8.25, 95%CI). CONCLUSIONS Preoperative polysomnography yields significant independent predictors of post-op complications in children with OSA. While AHI is a significant independent predictor, minimum O2 saturation on preop PSG appeared the strongest predictor when <80%. Patients with these risk factors, especially low O2 on PSG, warrant overnight observation with continuous pulse oximetry.