Three-dimensional echocardiography (3DE) provides volumetric measurements without geometric assumptions. Volume-rendered 3DE has been shown to be accurate for the measurement of right ventricular (RV) volumes in vitro and in animal studies; however, few data are available regarding its accuracy in patients. This study examined the accuracy of 3DE for quantitation of RV volumes and ejection fraction (EF) in patients, compared to magnetic resonance imaging (MRI) and radionuclide ventriculography (RNV). Twenty patients underwent MRI, gated equilibrium RNV, and 3DE using rotational acquisition from both the transesophageal and transthoracic approaches. RV volumes and EF were calculated from the 3DE data using multislice analysis (true Simpson's rule). RV volumes calculated by MRI (end-diastolic volume (EDV) 109.4 +/- 34.3 mls, end-systolic volume (ESV) 59.6 +/- 31.0 mls, and EF 47.7 +/- 17.1%) agreed closely with 3DE. For transesophageal echocardiography, EDV was 108.1 +/- 29.7 mls (r = 0.86, mean difference 1.3 +/- 17.8 mls); ESV was 62.5 +/- 23.8 mls (r = 0.85, mean difference 2.8 +/- 15.1 mls); and EF was 43.2 +/- 11.7% (r = 0.84, mean difference 4.5 +/- 9.7%). For transthoracic echocardiography, EDV was 107.7 +/- 27.5 mls (r = 0.85, mean difference 1.6 +/- 18.2 mls); ESV was 59.7 +/- 22.1 mls (r = 0.93, mean difference 3.2 +/- 19.6 mls); and EF was 45.2 +/- 11.5% (r = 0.86, mean difference 2.0 +/- 9.4%). There were close correlations, small mean differences and narrow limits of agreement between RNV-derived EF (43.4 +/- 12.1%) and both transesophageal (r = 0.95 mean difference 0.2 +/- 3.7%) and transthoracic 3DE (r = 0.95, mean difference 1.8 +/- 5.4%). Three-dimensional echocardiography is a promising new method of calculating RV volumes and EF, comparing well with MRI and RNV. The accuracy of transthoracic 3DE was comparable to that of the transesophageal approach. Three-dimensional echocardiography has the potential to be useful in the clinical assessment of RV disorders.