Accurate and reproducible quantification of tumor radioactivity by imaging requires definition of a region of interest (ROI) for the tumor. The use of a threshold for creating the tumor ROI based on tumor-to-background image contrast (image contrast) was examined. Quantification of iodine-131 in spheres in a phantom that simulated tumors in patients was investigated using planar imaging and geometric-mean and effective-point-source methods. Thresholds that provided the least quantitative error for spheres with different diameters (1-5 cm) and locations (0-11 cm deep in the body), 131I concentrations (0.037-3.2 MBq/ml), and sphere-to-background concentration ratios (1:0, 14:1 and 7:1) were investigated. The correlation between threshold and sphere image contrast was examined. The phantom study showed that an appropriate threshold value for quantification of tumor radioactivity could be determined using image contrast for a single view, provided that image contrast was >/=1.5. The error of quantification was less than 10% for spheres with high image contrast (>/=1.5) but was greater than 17% for spheres with low image contrast (<1.5). When image contrast-dependent thresholds were applied to patient studies, 131I concentrations determined by imaging were in good agreement with the concentrations determined by counting biopsy samples. Additionally, reproducibility was improved when compared with a visual boundary method. It is concluded that accurate and reproducible quantification of radioactivity in tumors is achievable using thresholds based on image contrast if image contrast is greater than or equal to 1.5. Optimal thresholds for quantification of tumor radioactivity were similar if image contrast was similar despite differing tumor diameters, locations and 131I concentrations. Under certain circumstances, the effective-point-source method was preferable to the geometric-mean method.