MRI is an established and accurate method of measuring left and right ventricular volumes by summing chamber areas in multiple contiguous slices. Acquisition time may be up to 45 min. We have estimated volumes with gradient echo imaging to test the accuracy of a more rapid method (total acquisition time 15 min) using a recognized echocardiographic algorithm. The results were compared with the spin echo method. We studied 20 patients (mean age 52 years, 15 male) within 6 months of anterior myocardial infarction and 20 normal subjects (mean age 40 years, 19 males). For the rapid method, cine acquisitions were made in the horizontal long axis plane and in two short axis planes which divided the long axis into three equal parts. Volume was calculated assuming the ventricle to be composed of a cylinder, a truncated cone and a cone. There was good agreement between the two methods at end diastole with a mean difference (+/- standard error, +/- 95% confidence interval for limits of agreement) of -3 ml (+/- 8.3, +/- 37%) for normal subjects and 1.5 ml (+/- 4.2, +/- 25%) for patients. Agreement was less good at end systole with mean difference of 12.1 (+/- 3.5, +/- 41%) for normal subjects and 25.7 (+/- 3.7, +/- 47%) for patients. The rapid method, therefore, significantly underestimated end systolic volume compared with the previous method. Rapid measurements of end diastolic volume are more accurate than those of end systolic volume and hence ejection fraction. Provided the potential error is recognized, the rapid technique can be used in routine clinical practice in both normal and abnormal ventricles.