The athymic (nude) mouse is a useful model for studying the biology and response to therapies of human tumors in vivo. A survey of recent literature revealed the use of 19 different formulas for determining the size of subcutaneous tumors grown as xenografts in nude mice (2 for determining tumor area, 3 for tumor diameter, and 14 for calculating tumor volume). We compared the volumes, areas, and diameters predicted by each of the 19 formulas with the actual weights of 50 tumors ranging from 0.46 to 22.0 g established in nude mice as xenografts from human cell lines. In addition to determining how well each formula predicted relative tumor size, we analyzed how well each formula estimated actual tumor mass. The ellipsoid volume formulas (pi/6 x L x W x H and 1/2 x L x W x H) were best for estimating tumor mass (r = 0.93), whereas measurements of diameter correlated poorly with tumor mass (r less than 0.66). Although determination of tumor area correlated well with tumor mass in small tumors (r = 0.89), correlations of area with tumor mass for large tumors were poor (r = 0.41). We conclude that determination of the ellipsoid volume from measurements of three axes consistently yields the most accurate estimations of both relative and actual tumor mass.