To determine if heat tolerance could be predicted from responses to exercise in temperature conditions, 51 young men performed 15 min of bench stepping at an average work load of 80 W at 23 degrees C. On the following day they attempted to perform 3 h of bench stepping at 40 W in heat (39.3 degrees C dry bulb, 30.3 degrees C wet bulb). Of these subjects, 4 were heat intolerant, judged by previous heat stroke episodes during field marches, 12 were heat acclimated, and 35 were unacclimated. The heat-intolerant subjects showed the highest heart rates (HR) and rectal temperatures (Tre) at 23 degrees C and in heat, and the acclimated subjects showed the lowest corresponding values. HR and Tre in each environment were combined into a single score, from 10, indicating the poorest responses, to 100, indicating the best responses. These scores at 23 degrees C when correlated with the corresponding scores in heat resulted in a linear correlation coefficient of r = 0.94 with a standard error of estimate of 8.6%. Scores of the heat-intolerant subjects were below 35, and those of the acclimated subjects were between 70 and 100. Thus heat tolerance can accurately be predicted for HR and Tre responses to exercise at room temperature.