1. The lowest water temperature in which different young adults could stabilize body temperature was found to vary from 32 degrees C to less than 12 degrees C, because of large differences in both total body insulation and metabolic heat production. 2. Total body insulation per unit surface area, in the coldest water allowing stability, was quite closely determined by mean subcutaneous fat thickness measured ultrasonically (r = 0.92), regardless of differences in distribution of this fat between men and women. 3. Reactive individuals developed high metabolic rates, and often rather high insulations in relation to fat thickness, which enabled them to stabilize their body temperatures in water more than 10 degrees C colder than was possible for less reactive individuals of similar fat thickness. 4. Measurements of heat flux, after stabilization in the coldest water possible, showed that the trunk was the main site of heat loss and that over half of the internal insulation there could be accounted for by subcutaneous fat; by contrast, fat could account for less than a third of higher insulations found in muscular parts of the limbs, and for less than 3% of very high insulations in the hands and feet. 5. After stabilization of body temperature at rest in the coldest possible water, exercise reduced internal insulation only in muscular parts of the limbs. Exercise also increased heat loss elsewhere by exposing skin of protected regions such as flexural surfaces of joints. During exercise total heat production increased rather more than heat loss in unreactive subjects, but less than loss in subjects whose heat production had already risen to a high level when they were at rest in cold water. 6. In warm (37 degrees C) water, tissue insulations were lower and much more uniform between subjects and between different body regions than in the cold. Even in the warm, however, insulations remained rather higher in fat than thin subjects, higher at rest than during exercise, and usually higher in the limbs than the upper trunk.