The aim of this study was to look at the influence of homelessness on acute medical admissions. A prospective case-controlled study was therefore performed on all homeless children admitted through the accident and emergency department over one year, comparing them with the next age matched admission from permanent housing. Assessments made were: whether homelessness or other social factors influenced the doctors' decision to admit; differences in severity of illness; length of stay; and use of primary care. The admitting doctors completed a semi-structured questionnaire during admission about social factors that influenced their decision to admit and graded the severity of the child's illness. The length of hospital stay was recorded. The family's social risk factors and accommodation were assessed at a home visit using a standardised questionnaire and by observation. Seventy homeless children were admitted. Social factors influenced the decision to admit in 77% of homeless children and 43% of controls. More of the homeless children were only mildly ill (33/70) than those from permanent housing (21/70), although three of the homeless children died of overwhelming infections compared with none of the controls. Among homeless families many were recent immigrants (44%). There was a marked increase in socioeconomic deprivation, in major life events in the previous year (median score 3 v 1), and in maternal depression (27% v 8%). Referral to the hospital was made by a general practitioner in only 5/50 (10%) of homeless compared with 18/50 (36%) of controls. Social factors were an important influence on the decision to admit in over three quarters of the homeless children and resulted in admission when less severely ill even when compared with admissions from an inner city population. Even though there was marked social deprivation among the homeless families, the decision to admit was based on vague criteria that need to be further refined.