Left ventricular (LV) volumes and ejection fraction can be obtained by applying Simpson's rule to multiple short-axis tomographic planes. A simpler method for determining LV volumes using the area-length equation is widely accepted and requires less time to acquire and analyze. Its accuracy, however, is questionable in deformed or asymmetrically contracting ventricles. This study compares biplane long-axis to serial short-axis computed LV volumes obtained by cine gradient-echo magnetic resonance imaging (MRI) in 2 distinct patient populations: (1) patients with global LV dysfunction, and (2) patients with regional LV dysfunction. A total of 114 patients were studied using both methods. Among 37 patients with global LV dysfunction, there was no statistically significant difference between methods (long axis vs short axis) for determining LV end-diastolic volume (203 +/- 91 vs 201 +/- 90 ml), end-systolic volume (142 +/- 81 vs 141 +/- 82 ml), and ejection fraction (33 +/- 12 vs 33 +/- 13%). However, in the 77 patients with regional dysfunction, LV end-diastolic volume was statistically slightly higher when obtained using the long-axis approach (157 +/- 53 vs 152 +/- 51 ml; p=0.004). Otherwise, end-systolic volume (97 +/- 49 vs 95 +/- 49 ml) and ejection fraction (40 +/- 13 vs 40 +/- 13%) were similar (p=NS). The correlation between LV volumes and ejection fractions for both groups was excellent (r >0.91). Thus, in this study group, biplane long-axis and serial short-axis computed LV volumes and ejection fractions were similar in patients with global or regional LV dysfunction. In critically ill patients unable to complete a comprehensive MRI examination, the biplane long-axis-derived volumes provide adequate data.