A new automated edge detection program has been developed to estimate left ventricular mass from single photon emission computed tomographic (SPECT) 201Tl images and 14 dogs were studied. Six of the 14 dogs underwent imaging before and 5 hr after coronary artery occlusion with a closed-chest technique. True left ventricular mass was determined at time of killing within 1 hr of the last 201Tl study. Left ventricular mass determined by tomography correlated well with autopsy left ventricular mass (r = 0.94; p less than 0.001, s.e.e. = 5.9 g) over a range of 62-156 g. The intraobserver variation between repeated measurements of the same SPECT study yielded an r = 0.99; p less than 0.0001; s.e.e. = 2.3 g. The reproducibility of the mass determination was assessed in four animals with two studies performed 10-14 days apart. The estimate of left ventricular mass from the two studies were highly correlated (r = 0.98; p less than 0.001) with a mean absolute difference of 4 g (3.3%). In the six dogs with a control and postinfarct study the mean total left ventricular mass by 201Tl tomography varied by less than 3.8% (r = 0.89; p less than 0.001). In conclusion, tomographic imaging with 201Tl can define left ventricular mass accurately and reproducibly in the dog model. The ability of this 201Tl tomographic left ventricular mass program to measure both normal and infarcted tissue accurately suggests the possibility of documenting interventions designed to alter left ventricular mass and of sizing acute infarcts and assessing interventions that may alter acute infarct size.