This report, which is one of a series, is concerned with socio-demographic correlates of self-referral, and the extent to which the observed associations can be explained by variations in symptom prevalence and severity. Social class, marital status, employment status, and distance from health centre, all show small but significant associations with self-referral. For social class the effect appears to be mediated by symptoms. People in social classes IV and V, women who are widowed, divorced or separated, and people who live near the health centre are more likely to visit their general practitioner.