200 Background: Baseline pain has been shown to be a significant predictor of overall survival (OS) in mCRPC pts. The objectives of this analysis were to identify predictors of OA use at 3-months after initiating frontline chemotherapy, and to determine if OA use at 3-months predicts overall OS in mCRPC pts. Methods: Data from CALGB 90401, a phase III trial of 1,050 mCRPC pts randomized with equal probability to receive docetaxel, prednisone with either bevacizumab or placebo was analyzed. OA use was collected at baseline and at 3-month follow-up. OA use was classified as: pts not on OA at baseline and at 3-months (OA1); OA at baseline but not on at 3-months (OA2); No OA at baseline but on OA at 3-months (OA3); and OA at both time points (OA4). The logistic regression method was utilized to identify predictors of OA use at 3-months. Furthermore, the proportional hazards model was used to assess the prognostic significance of OA use in predicting OS adjusting for stratification factors. Results: 836 pts had complete OA data. The overall distribution of pts with the associated median survival times is presented in the table. There was a linear trend between OA use and overall survival (log-rank p-value<0.0001). The multivariable hazard ratio (HR) for death was 1.34 (95% CI=1.12-1.61) comparing OA4 vs. OA1. In multivariable logistic regression model, statistically significant predictors of OA at 3-months were OA use at baseline, hemoglobin, and alkaline phosphatase. Conclusions: The strongest predictor of OA use at 3-months was use at baseline. In addition, this analysis shows that OA use after initiating chemotherapy is a statistically significant predictor of overall survival in mCRPC pts. These results require prospective validation. [Table: see text]