Sudden infant death syndrome (SIDS) death rates show substantial seasonal variation. Analysis of this variation shows that it is due to postnatal factors. Possible variables are temperature, infection, pollution, and the infants' interaction with the social environment. In England and Wales the temporal pattern of pollution levels and the geographical variation of death rates do not suggest pollution is generally an important factor. Social environment appears the most important. One in three high risk infants are in families with multiple social problems compared with 9% of controls. Further studies of why some infants are at greatly increased risk might identify crucial adverse environmental factors and indicate how some deaths could be prevented.