The heterogeneity of response to hyperthermia of cells taken from different regions of tumors was tested in a model tumor system (RIF-1) in the mouse and in specimens from spontaneous tumors taken from dogs and humans at the time of surgical resection. Cell survival was assayed by clonogenic survival in the murine tumor and by dansyl lysine staining in tumors from all three species. Using survival as an endpoint, it was found that the extent of heterogeneity depended on the temperature to which the tumor was heated and the duration of exposure. By increasing either of these factors, the coefficient of variation was increased. The large heterogeneity seen after in vivo heating could not be explained entirely by inhomogeneous heating within the tumor as evidenced by temperature mapping. It is concluded that other microenvironmental factors such as blood flow, pH, O2, and nutrient supply may cause variations in the heat response of the tumor cells in vivo. Little, if any, evidence of cellular heterogeneity was evident for all three species when comparisons were made between samples of 100-200 mg. The canine and human tumors were considerably more heat resistant when dansyl lysine was used as an endpoint. In the RIF-1 tumors, heterogeneity of heat response was greater after in vitro heating than after in vivo heating when small biopsy samples (10-20 mg) were taken, suggesting that some cellular heterogeneity was present.