A study of heat stress in underground miners was conducted to evaluate several instruments commonly used to determine heat stress. The study group consisted of 15 experienced acclimatized miners, mean age 33.2 years, working at seven sites in an underground mine in which a chiller and vent bag were used for climate control. Environmental heat loads were monitored at 30 minute intervals over a work shift using sling psychrometers; dry bulb, natural wet bulb (NWBT), globe, botsball, and wet kata thermometers; direct reading electronic wet bulb globe temperature (WBGT) meters; and vane anemometers. Oxygen consumption, standard recovery heart rates, and oral temperature were also monitored. Pre and post shift body weights were measured to assess the extent of dehydration. A time motion study was done on each subject in order to calculate metabolic heat production rates. The data were used to calculate heat stress indices. The mean dry bulb temperature was 26.5 degrees- C, a value not significantly different from the wet bulb temperature. The mean wet bulb depression varied from 0.8 to 1.3 degrees-C, indicating a relative humidity above 90 percent. Air velocities at the work sites ranged from 0.2 to 1.7 meters per second. None of the NWBTs exceeded the American Conference of Governmental Industrial Hygienists threshold limit values for light work, but three NWBTs exceeded the limits for moderate work. None of the oral temperatures exceeded 37.6 degrees-C; however, recovery heart rates frequently exceeded 125 beats per minute when heavy work was done. The oral temperatures and heart rates correlated poorly with the heat stress indices. Five subjects had weight losses exceeding 1.5 percent of their body weight. The mean metabolic heat production rates ranged from 175 to 500 watts. Heat stress indices calculated from the environmental measurements correlated well with each other. The authors conclude that the botsball thermometer is the best instrument for assessing heat stress in underground mines of this type because it gives the most reliable estimate of WBGT and is easy to use.