OBJECTIVES: Cost of treatment for patients with the same diagnosis can vary enormously due to differences in comorbidities, practice patterns, and outcomes of care. We sought to incorporate this variability in a model projecting the results of clinical trials to community practice. METHODS: We modeled an episode of care for persons hospitalized due to acute decompensated heart failure and urgently treated with either nesiritide or dobutamine. Patient characteristics and probability of significant clinical events (cardiac arrest, sustained and non-sustained ventricular tachycardia, hypotension, vomiting, readmission and death during follow-up) were based on pooled analysis of two completed clinical trials. The cost of a hospital admission was derived from a subset of records from the 1997 HCUP hospital database for discharges with similar demographic and clinical features (n=57,223). Regressions were estimated for equations explaining (1) hospital length of stay as a function of patient attributes and specified clinical events; (2) the cost of the admission as a function of patient atributes, clinical events and predicted length of stay (LOS). For each of 5000 simulated patients, the model first stochastically generates new sets of regression parameters using the means and standard deviations of the original parameter estimates. Next the model predicts patient demographic characteristics and incidence of clinical events. The vector of patient attributes is applied to the vectors of regression parameters to predict LOS and then cost as a function of predicted LOS. RESULTS: This approach preserved distributional characteristics of the original HCUP data (e.g. model predicted cost of admission vs HCUP: mean 14,807 vs 14,666; skew 2.94 vs 3.16; kurtosis 10.03 vs 11.43) while enabling us to differentiate study drugs based on incidence of clinical events. CONCLUSIONS: The model yields robust estimates of cost. Confidence intervals surrounding point estimates offer decision-makers a reliable basis for assessing potential financial impact and uncertainty surrounding adoption of the treatment intervention.