OBJECTIVE: Simulation studies can be used to estimate or forecast drug effectiveness in a representative sample by translating a survival benefit established in a clinical trial. The methods of this translation have not been described. The present study describes how the translation is operationalized using a combination of clinical trial and observational data. We simulate the population impact of docetaxel (D), the first chemotherapy agent to demonstrate a survival benefit in patients with hormone refractory prostate cancer (HRPC). The survival benefit of D was established in the TAX327 trial however it is unclear how the benefit may translate to a heterogeneous population of patients. METHODS: A combination of TAX327 trial data and SEER-Medicare (SM) data were used. In pts age 69+ and randomized to D (every 3 weeks, D3P) or mitoxantrone (M), trial data showed a survival benefit for D. Accordingly, 3,515 SM pts age 69+, diagnosed with M1 PC between 1994 and 2002, and receiving only androgen deprivation therapy (ADT) were selected. Graphical plots and likelihood ratio tests were used to identify the best-fitting parametric survival functions for D3P, M, and SM patients. Candidate (selected) curves included: gamma, log-normal, weibull, exponential, and polynomial (log-logistic) functions. The survival benefit function from the trial was calculated using parametric curves fitted to D3P and M data and subsequently imposed on the SM survival curve. The simulated benefit was assessed in 3 scenarios depending on the assumed time of chemotherapy initiation: at diagnosis of M1 PC, at 12 mos post-diagnosis, and at 24 mos post-diagnosis. The simulated benefit in the SM sample was examined following adjustments for sample heterogeneity due to demographic, clinical, and ecologic factors. RESULTS: We find that the simulated survival benefit is sensitive to adjustments for sample heterogeneity and also is sensitive to assumptions regarding the time of chemotherapy initiation.