Adult male CD-1 mice and CD rats were used to determine LD50/24 h lethality rates from exposure to 2450-MHz circularly polarized microwaves. Groups of 16 mice or six rats were exposed in each of 32 combinations of nominal power density (10, 25, 50 or 75 mW cm-2), exposure duration (1 or 4 h), and environmental temperature (20 or 30 degrees C) and relative humidity (35 or 80%). An analysis of variance probit model was used to determine the influence each variable had on the probability of death. Significant factors in lethality were nominal power density, exposure duration and environmental temperature, but not environmental relative humidity. The estimated power density (mW cm-2) required to kill 50% of the animals in 24 h is halved when the environmental temperature is increased from 20 to 30 degrees C. Similarly, only 20-25% of the power density is required when the exposure duration is increased from 1 to 4 h. The use of nominal power density as a predictor of the probability of death was more efficient than specific absorption rate estimated experimentally by twin-well calorimetry. The exposure of one mouse at a time, instead of 16, did not alter the predicted death rate.