Objectives Three-dimensional echocardiography (3DE) calculates left ventricular volumes (LVV) and ejection fraction (EF) without geometric assumptions, but prolonged analysis time limits its routine use. This study was designed to validate a modified 3DE method for rapid and accurate LVV and EF calculation compared with magnetic resonance imaging (MRI). Methods Forty subjects included 15 normal volunteers (group A) and 25 patients with segmental wall motion abnormalities and global hypokinesis caused by ischemic heart disease (group B) who underwent 3DE with precordial rotational acquisition technique (2-degree interval with electrocardiographic and respiratory gating) and MRI at 0.5 T, electrocardiogram (ECG)-triggered multislice multiphase T1-weighted fast field echo. End-diastolic and end-systolic LVV and EF were calculated from both techniques with Simpson’s rule by manual endocardial tracing of equidistant parallel left ventricular short-axis slices. Slicing from the 3DE data sets were done by both 2.9-mm slice thickness (method 3DE-A) and by 8 equidistant short-axis slices (method 3DE-B); for MRI analysis, 9-mm slice thickness was used. Results Analysis time required for manual endocardial tracing of end-diastolic and end-systolic short-axis slices was 10 minutes for the 3DE-B method compared with 40 minutes by the 3DE-A method. For all 40 subjects the mean ± SD of end-diastolic LVV (mL) were 181 ± 76, 179 ± 73, and 182 ± 76; for end-systolic LVV (mL), 120 ± 76, 120 ± 75, and 122 ± 77; and for EF (%), 39 ± 18, 38 ± 18, and 38 ± 18 for MRI, 3DE-A, and 3DE-B methods, respectively. The differences between 3DE-A and 3DE-B with MRI for calculating end-diastolic and end-systolic LVV and EF were not significant for the whole group of subjects as well as for the subgroups. The 3DE-B method had excellent correlation and close limits of agreement with MRI for calculating end-diastolic and end-systolic LVV and EF: r = 0.98 (–1.3 ± 26.6), 0.99 (–1.6 ± 21.2), and 0.99 (0.2 ± 5.2), respectively. The correlation between 3DE-A and MRI were r = 0.97, 0.98, and 0.98, and the limits of agreement were –1.4 ± 36, –0.6 ± 26, and 0.6 ± 8 for calculating end-diastolic and end-systolic LVV and EF, respectively. In addition, excellent correlation and close limits of agreement between 3DE-A and 3DE-B with MRI for LVV and EF calculation was also found for the subgroups. Intraobserver and interobserver variability (SEE) of MRI for calculating end-diastolic and end-systolic LVV and EF were 6.3, 4.7, and 2.1; and 13.6, 11.5, and 4.7; respectively, whereas that for 3DE-B were 3.1, 4.4, and 2.2; and 6.2, 3.8, and 3.6; respectively. Comparable observer variability was also found for the A and B subgroups. Conclusions The 3DE-A and 3DE-B methods have excellent correlation and close limits of agreement with MRI for calculating LVV and EF in both normal subjects and cardiac patients. The 3DE-B method by paraplane analysis with 8 equidistant short-axis slices has observer variability similar to MRI and reduces the 3DE analysis time to 10 minutes, therefore offering a rapid, reproducible, and accurate method for LVV and EF calculation. (Am Heart J 1999;137:134-43.)