Cardiac volumes including left ventricular mass were determined by magnetic resonance imaging (MRI) in 30 patients using a contiguous multislice technique. Eleven patients were studied on two different days. Intra- as well as interobserver and interstudy variation were evaluated. For volume measurements the standard deviation of differences (SDD) was rather constant, independent of the size of the measured volume indicating lower percentual variability in measuring large volumes. Lowest variability was noticed for left ventricular myocardial mass estimates between observers as well as between studies. In 13 patients left ventricular mass was estimated by this technique as well as by a method using only one long axis and four short axis slices and by two-dimensional echocardiography. Although highly correlated, limits of agreement between the methods were too wide to accept interchangeable clinical use of the methods.