Details of room temperature, clothing, and bedding used by night and by day and in winter and in summer were recorded for 649 babies aged 8 to 26 weeks. Room temperature at night was significantly related to outside temperature and duration of heating. Total insulation was significantly related to outside temperature and to minimum room temperature, but there was wide variation in insulation at the same room temperature. High levels of insulation for a given room temperature were found particularly at night and in winter, and were associated with the use of thick or doubled duvets and with swaddling. At least half the babies threw off some or all of their bedding at night, and at least a quarter sweated. Younger mothers and mothers in the lower social groups put more bedclothes over their babies, and the latter also kept their rooms warmer. Many mothers kept their babies warmer during infections.There has recently been interest in the thermal environment of sleeping babies, arising mainly from suggestions that some cases of sudden infant death syndrome (SIDS)'" and haemorrhagic shock and encephalopathy syndrome (HSES)5 may be linked with overheating. A difficulty in evaluating this hypothesis has been the lack of control data. There have been four published surveys, from Exeter,6 Nottingham,7 Leicester,8 and Dunedin (New Zealand),9 of how mothers keep their babies warm.We present the findings of a population survey of the thermal environment of 649 sleeping babies by day and night in summer and in winter. weeks (n=238), or 26 weeks old (n=207). The interviewer recorded full details of clothing and bedding used for the baby's sleep the previous night and for sleep on the day of the visit, inspecting all items to check their type and material. She also recorded the social background of the family, the gestation and birth weight of the baby, the type and duration of heating, the ventilation of the room, whether the baby sweated while sleeping, and what adjustments the mother made if the baby had an infection.The ambient temperature of the baby's sleeping place was measured, for day by a recording at the time of the visit, for night by a maximumminimum thermometer left near the cot and collected next morning (temperature recordings were not available for three babies). Outside temperatures for the same times were obtained from the Meteorological Office. The study continued through all months of the year except August. The data were analysed separately for the 326 mothers who were interviewed in summer and the 323 who were interviewed in winter, summer being defined as the period of British summer time. and were approached for interview. Of these 329 (33-6%) declined to participate or were not available. The social class distribution (by father's occupa-