Induction of unrepairable DNA damage, accumulation of misrepaired DNA damage, and generation of imbalances in competing biochemical and/or metabolic processes have been proposed to explain the relationship between radiation-induced DNA damage and cell lethality. Theoretically, the temperature dependence of the critical DNA repair process(es) should be 1) either independent of or identical to the temperature dependence of cell killing if the first two hypotheses are correct, and 2) different if the third hypothesis is correct. To test this, exponentially growing rat 9L brain tumor cells were left at 37 degrees C or equilibrated for 3-14 h at 20 degrees C before irradiation. Cells were irradiated and allowed to repair at either 20 degrees C or 37 degrees C. Alternatively, the cells were irradiated at one of these temperatures and immediately shifted to the other temperature for repair. DNA damage was assessed by the alkaline elution technique; cell kill was assessed by a clonogenic assay. 9L cells maintained at 20 degrees C or 37 degrees C sustained the same amount of DNA damage as measured by alkaline elution. DNA repair instantaneously assumed the rate characteristic of the postirradiation temperature. For 9L cells equilibrated, irradiated, and repaired at 20 degrees C, the half-time of the fast phase of the DNA repair decreased by a factor of approximately 2 and the half-time of the slow phase decreased by a factor of approximately 5 over that measured in cells incubated, irradiated and repaired at 37 degrees C. Although the rate of DNA repair decreased substantially at 20 degrees C, the survival of 9L cells that were equilibrated and irradiated at 20 degrees C was greater (p less than 10(-4)) than those incubated and irradiated at 37 degrees C, when assayed by an immediate plating protocol. In addition, the survival of 9L cells equilibrated and irradiated at 20 degrees C and then shifted to 37 degrees C immediately after irradiation was greater (p less than 10(-2)) than that obtained with any other delayed plating protocol. Thus, the temperature dependence of the DNA repair processes measured by alkaline elution was different from the temperature dependence of cell killing measured either by an immediate or delayed plating protocol. These data support the hypothesis that many irradiated 9L tumor cells die because of imbalances in sets of competing biochemical and/or metabolic processes.