BACKGROUND: The goal was to assess the incremental diagnostic value of thallium single-photon emission computed tomographic (SPECT) imaging and lung/heart ratio (LHR) over other clinical and exercise electrocardiographic (ECG) variables concerning the presence and extent (three vessel/left main) of coronary artery disease. METHODS AND RESULTS: Multivariable logistic regression analysis that used an incremental study design was applied to clinical, exercise test, SPECT, LHR, and catheterization data from 323 patients with suspected coronary disease. The following variables were evaluated as predictors of presence and extent of disease: clinical (age, sex, symptoms, diabetes, and smoking), exercise ECG (millimeters of ST segment depression, slope of ST segment depression, peak heart rate, change in systolic blood pressure, and metabolic equivalents of the task), and thallium (defect reversibility and intensity of hypoperfusion and LHR). Discrimination and incremental value were assessed by receiver operating characteristic (ROC) curve analysis. SPECT imaging (segment hypoperfusion score) was an independent predictor of both presence and extent of disease (p < 0.0001) irrespective of the percent stenosis criterion used. However, it added significant incremental diagnostic information over clinical and exercise ECG data only concerning presence (e.g., ROC curve areas: presence of disease-clinical plus exercise ECG = 83 +/- 2 vs clinical, exercise ECG, plus SPECT = 87 +/- 2, p < 0.001; extent of disease-clinical plus exercise ECG = 83 + 4 vs clinical, exercise ECG, plus SPECT = 85 +/- 3, p = 0.11). Thallium LHR was an independent predictor of both presence and extent of disease (p < 0.05), but the incremental information added to SPECT data did not reach statistical significance (e.g., ROC curve areas: presence of disease--SPECT = 87 +/- 2 vs SPECT plus LHR = 88 +/- 2, p = 0.24; disease extent--SPECT = 85 +/- 2 vs SPECT plus LHR = 86 +/- 2, p = 0.24). However, when combined, LHR and SPECT data had significant incremental value over clinical and exercise ECG data alone concerning extent of disease (e.g., ROC curve areas: clinical plus exercise ECG = 83 +/- 4 vs clinical, exercise ECG, SPECT and LHR = 86 +/- 3, p = 0.04). CONCLUSION: SPECT thallium imaging variables are independent predictors of both presence and extent of coronary disease. However, they have significant incremental value over clinical and exercise ECG data concerning only presence of disease. Thallium LHR is an independent predictor of both presence and extent of disease but adds only a small insignificant amount of incremental diagnostic information over SPECT data. However, there was significant incremental value to the addition of LHR to SPECT data concerning extent of disease.