Ultrafast computed tomography has been reported to be an accurate method of measuring left ventricular mass in dogs. To assess the interstudy, intraobserver and interobserver variability of left ventricular myocardial mass measurements in humans, left ventricular myocardial volume was measured three times within 24 h in 16 patients with ischemic heart disease. The mean percent difference of the mean of the three studies performed was -0.01 +/- 1.4% (range -2.9% to 3.6%). The regression analysis for the intraobserver variability at baseline was: Y = -4.33 + 1.03X; r = 0.99, SEE = 3.5 ml. The mean percent difference of the mean of the two sets of measurements performed by two independent observers was 0.28 +/- 2.1% (range -4.35% to 4.35%). The interobserver variability excluding papillary muscles at baseline study was: Y = -4.34 + 1.06X; r = 0.99, SEE = 1.5 ml. The regression analysis with versus without papillary muscles showed: Y = -8.72 + 0.97X; r = 0.96, SEE = 2.6 ml. Regression analysis to assess the variability of 24-h studies at end-systole versus end-diastole revealed: Y = 3.07 + 0.94X; r = 0.97, SEE = 1.8 ml. In conclusion, ultrafast computed tomography is a minimally invasive technique, with very low interstudy, intraobserver and interobserver variability for left ventricular myocardial volume and mass determinations in serial studies.