In a case-control study of factors contributing to hypothermia all fourteen patients (mean age 80 years) admitted to hospital with hypothermia after being found ill indoors also had some other serious illness. They were more likely than control patients to have been alone when taken ill (93 vs 39% of controls), to live alone (86 vs 43%), and to have been found on the floor (79 vs 14%). They were less likely to have been wearing more than indoor clothing (0 vs 50%), or to have had heating on when found (50 vs 89%), but 93% of patients in both groups had heating available. Healthy young adult volunteers who lay immobile on the ground in air at 5 degrees C lightly clothed cooled progressively by 0.57 degrees C (SD 0.32) (rectal T degrees) in 90 min despite doubling of metabolic rate. With better insulation in bed, core temperature stabilised within 90 min, and when they were in an armchair it fell slowly, with no increase in metabolic rate in either case. The findings suggest that hypothermia indoors resulted largely from collapse due to illness when the patient was alone lightly clothed and not in bed. Eight hypothermic patients found outside (in December and January) were younger (mean age 60 years) than the fourteen found indoors; six of these were chronic alcoholics or acutely intoxicated, and six lacked, or had wandered from, a fixed home.