In this study a new method for assessment of perfusion defects (PDs) derived from myocardial perfusion tomograms was evaluated in patients treated with thrombolytic therapy. Using global constraints and dynamic programming, a model-based delineation algorithm defined myocardial borders, the basal plane and absolute and relative PD size in 49 thallium-201 chloride (201TL CL) and 60 technetium-99m methoxyisobutylisonitrile (99mTc-MIBI) tomograms. Tomographic (single-photon emission tomography: SPET) and planar quantification of PDs was compared to enzymatic infarct size as well as to global (LVEF) and regional ventricular function (RWM) obtained by contrast angiography. The algorithm delineated the myocardium and the valve plane in most cases, even when large PDs were present. Manual correction of the automatic delineation of the basal plane was necessary in less than 20% of the studies. Using 201Tl Cl, LVEF correlated better with tomographic PD (r = -0.67) than with planar PD (r = -0.54). Comparing planar to tomographic imaging using 99mTc-MIBI, a higher correlation with enzymatic infarct size (r = 0.73 vs 0.57) and with global ventricular function (r = 0.64 vs -0.52) was found when tomographic techniques were used. No close correlation between PD and RWM was found. The beneficial effect of thrombolysis was shown by a significant difference of PD in patients with open versus occluded infarct-related vessels. It can be concluded that this new automated algorithm for quantification of SPET perfusion defect size provides a useful tool in evaluating thrombolytic therapy.