We investigated tumor detectability of bronchial carcinoma using 201Tl and 99mTc-MIBI with planar scintigraphy and SPECT. We studied 30 tumor patients and 7 patients with coronary artery disease in a clinical phase III trial. The lung was partitioned into 6 areas, which were read independently, to provide a sufficiently large sample of tumor-free reference regions. We calculated the statistical power for the comparison of sensitivities at a given specificity of 95% (TPF-test) to demonstrate that the subdivision of the lung and the application of the bivariate bi-normal ROC model allows an objective assessment of diagnostic performance even for the small sample size of our study. There were no significant differences between 201Tl and 99mTc-MIBI for both observers; therefore, no advantage of 99mTc-MIBI over 201Tl for tumor scintigraphy could be demonstrated. SPECT was significantly superior to planar scintigraphy and should be preferred for tumor detection in the thorax.