The normal distribution of myocardial tracer activity is different in attenuation-corrected images compared with uncorrected images. We therefore postulated that quantitation of attenuation-corrected thallium 201 images with direct comparison to a database of healthy subjects could improve detection of coronary artery stenoses. In 49 patients with angiographic evidence of coronary artery disease and 69 patients with a less than 5% likelihood of coronary artery disease, tomographic Tl-201 myocardial imaging was performed by means of a triple-headed camera with fan-beam collimators and the images were processed with attenuation correction, with attenuation and Compton scatter correction, and without correction. Images were subjectively interpreted in a blinded manner. From the group of 69 patients with a low likelihood of coronary disease, scintigraphic data from the first 20 men and the first 20 women were used to generate normal reference ranges for each of the 3 image-processing methods. Data from the 49 patients with angiographic coronary disease and from the remaining 29 patients with a low likelihood of coronary disease were then analyzed quantitatively in comparison to the gender-matched normal databases. With visual interpretation, attenuation-corrected images yielded an improved detection rate compared with uncorrected images (79% vs 46%, P = .008) for 70% or greater left anterior descending coronary artery stenoses. Attenuation-corrected images also provided an improved normalcy rate for the right coronary artery territory (91% vs 75%, P = .006) and an improved normalcy rate when the images were analyzed overall by patient (88% vs 74%, P = .009). With quantitative analysis, attenuation correction yielded an improved detection rate for patients with a significant coronary artery stenosis compared with uncorrected images (90% vs 71%, P = .016). Defect extent was also increased by attenuation correction. Compared with attenuation correction alone, the addition of scatter correction yielded no incremental improvement in detection rate or normalcy rate. Attenuation-scatter correction did produce a nonsignificant trend toward detection of more of the total number of stenotic arteries compared with uncorrected images (60 vs 48 of 79 arteries). With the use of a triple-headed camera with fan-beam collimation, visual analysis of attenuation-corrected Tl-201 images improved detection of left anterior descending coronary artery disease without loss of detection in other coronary territories. Furthermore, quantitative analysis of attenuation-corrected Tl-201 images improved the overall detection rate for coronary artery disease in patients without compromise of the normalcy rate.