To identify the transthoracic echo-Doppler (TTE) variables most predictive of significant mitral regurgitation (MR) of mechanical prosthetic valves, TTE and trans-esophageal echo (TEE) studies were independently reviewed in 57 patients (mean age [+/-SD] 59+/-12.5 years) undergoing both studies within 2+/-3 days. Several 2-dimensional and Doppler hemodynamic variables from the TTE studies were derived. Prosthetic MR was significant (moderate or severe) by TEE in 20 patients, whereas mild or no MR was seen in 37 patients. The best univariate predictors of significant MR by TTE were peak velocity of mitral inflow, mean gradient, tricuspid regurgitation velocity, isovolumic relaxation time, and ratio of time velocity integral of mitral inflow to time velocity integral in the left ventricular outflow (TVI(MV)/TVI(LVO)). Peak mitral velocity and TVI(MV)/TVI(LVO) were the best predictors of significant MR and performed similarly (area under the receiver-operating characteristic curve: 0.97 for both). A peak velocity of > or =1.9 m/s was 90% sensitive and 89% specific for significant prosthetic MR, whereas a TVI(MV)/TVI(LVO) > or =2.5 had a sensitivity and specificity of 89% and 91%, respectively. A decision tree was constructed to assess the conditional probabilities of having significant MR given all the possible outcomes of the 2 best predictors. None of the patients with peak velocity < 1.9 m/s and TVI(MV)/TVI(LVO) <2.5 by TTE had significant MR. Conversely, all patients with peak velocity > or =1.9 m/s and TVI(MV)/TVI(LVO) > or =2.5 had significant MR. The use of more complex algorithms did not further improve the results. Thus, measurements of hemodynamic Doppler variables on TTE examination can accurately identify a large number of patients without significant prosthetic MR, thereby reducing the need for further investigation with TEE.