Right ventricular ejection fraction (RVEF) is used clinically as an index of right ventricular (RV) pump function. Clinical measurements of RVEF are complicated by the need for complex imaging equipment to compute RV volumes. Recently, the use of thermodilution (TD) methods have been suggested as a simplified means to measure RVEF (RVEFTD) in patients using rapid response thermistors. Validation, however, by comparison of RVEFTD and other methods in vivo, is difficult. Accordingly, thermodilution derived EF measurements (EFTD) were compared to known values using an in vitro system, with known ejection fractions (EF) set from 17-78% and stroke rates varying independently from 50-100 strokes/min. EFTD was computed by fitting the downslope of the TD curve to a monoexponential function and computing the time constant of thermal decay. A significant correlation existed between EFTD and actual EF over the entire study (r = 0.96, p less than 0.001). Bias analysis showed that the points were within a 95% confidence interval of +/- 12%. Multivariate analysis showed that stroke rate did not significantly affect TD measurements (r = 0.03, p greater than 0.7). This study demonstrates that TD accurately predicts EF using an in vitro system and appears to be independent of stroke rate. Thus, TD methods may provide an accurate, simple and reliable means to serially measure RVEF in the clinical setting.