To determine whether a new approach to interpretation could improve the accuracy of thallium-201 single photon emission computed tomography (SPECT) for detection of left anterior descending coronary artery disease in patients with left bundle branch block, 69 patients were evaluated. Forty-four had angiographically proved coronary artery disease; the remaining 25 were considered to have a "low" (mean 13.5 +/- 6.4%, range 3.4% to 24.9%) likelihood of disease before thallium-201 scintigraphy. The conventional scintigraphic criterion for detection of left anterior descending artery disease (septal, anterior or apical defects) was compared with a new criterion that required the apex to be abnormal to indicate left anterior descending disease. The normalcy rates in the low likelihood patient group were significantly improved by using the new approach, from 16% to 80% (p less than 0.0001) by visual analysis and from 24% to 64% (p = 0.003) by quantitative SPECT polar map analysis. The sensitivity for left anterior descending disease was similar for the conventional and the new method by visual (100% vs. 94%) and quantitative (100% vs. 83%) analyses. In contrast, the specificity was significantly improved by using the new approach, from 14% to 79% (p = 0.0006) by visual analysis and 14% to 64% (p = 0.007) by quantitative analysis. In conclusion, septal and anterior thallium-201 SPECT defects are common in patients with left bundle branch block without coronary artery disease, resulting in low specificity for left anterior descending artery disease. The normalcy rates and accuracy for detection of left anterior descending coronary artery disease were significantly better when an apical defect was used as the criterion for disease.