Background: TROPIC evaluated the efficacy and safety of the novel taxane cabazitaxel in men with mCRPC previously treated with docetaxel. Median OS was significantly improved, as previously reported (12.7 months in mitoxantrone arm vs 15.1 months in cabazitaxel arm, HR=0.72 [0.61 - 0.84], p<0.0001- updated OS results). Median OS is the most useful descriptive statistic for physicians and patients as it reflects a point estimate in time by which 50% patients may survive regardless of disease status or progression. It avoids assumptions on long-term survival beyond the follow-up period of the clinical trial. Payers however are interested in making a coverage and reimbursement decisions based on the overall therapeutic benefit relative to its risk, and the expected impact on healthcare expenditures. Analyses such as cost-effectiveness analysis therefore require the estimation of mean OS. Methods: Mean OS is only observable when the last patient has died. Its estimation can be derived via an extrapolation of the trial Kaplan-Meier curve using a survival function. Several parametric distributions (exponential, weibull, lognormal, loglogistic and Gompertz) were tested. The parametric distribution that best fitted the trial Kaplan-Meier curves was selected using the Akaike's Information criteria (AIC), the Bayesian Information Criteria (BIC) and graphical method to evaluate the goodness of fit of the distributions. Mean OS from the best fitted model was generated to support payer decision making. Results: Using AIC/BIC and graphical method, the Weibull survival function, S(t)=exp(-l t^s) where l is a scale parameter and s a shape parameter, was selected as the distribution that best fitted the TROPIC data. Results of the estimated mean survival assuming a Weibull function are described in the table. Conclusions: Assuming a Weibull distribution, mean OS is estimated at 14.5 months in mitoxantrone arm vs 18.5 months in cabazitaxel arm, leading to 4 months OS difference in favour of cabazitaxel.