To determine the accuracy of echocardiographic left ventricular (LV) dimension and mass measurements for detection and quantification of LV hypertrophy, results of blindly read antemortem echocardiograms were compared with LV mass measurements made at necropsy in 55 patients. LV mass was calculated using M-mode LV measurements by Penn and American Society of Echocardiography (ASE) conventions and cube function and volume correction formulas in 52 patients. Penn-cube LV mass correlated closely with necropsy LV mass (r = 0.92, p less than 0.001) and overestimated it by only 6%; sensitivity in 18 patients with LV hypertrophy (necropsy LV mass more than 215 g) was 100% (18 of 18 patients) and specificity was 86% (29 of 34 patients). ASE-cube LV mass correlated similarly to necropsy LV mass (r = 0.90, p less than 0.001), but systematically overestimated it (by a mean of 25%); the overestimation could be corrected by the equation: LV mass = 0.80 (ASE-cube LV mass) + 0.6 g. Use of ASE measurements in the volume correction formula systematically underestimated necropsy LV mass (by a mean of 30%). In a subset of 9 patients, 3 of whom had technically inadequate M-mode echocardiograms, 2-dimensional echocardiographic (echo) LV mass by 2 methods was also significantly related to necropsy LV mass (r = 0.68, p less than 0.05 and r = 0.82, p less than 0.01). Among other indexes of LV anatomy, only measurement of myocardial cross-sectional area was acceptably accurate for quantitation of LV mass (r = 0.80, p less than 0.001) or diagnosis of LV hypertrophy (sensitivity = 72%, specificity = 94%).(ABSTRACT TRUNCATED AT 250 WORDS)