An examination is made of various hypotheses which have been developed to account for the frequently observed inverse relationship between social class and psychological disturbance. Two of these explanations--the social selection and social causation hypotheses--are tested in a study designed to enable an assessment of their relative power to be made. Quasi-random samples of 200 Indian-born, 200 Pakistani-born and 240 native-born residents of England were interviewed and measures of social class, psychological symptoms and other relevant variables made. The results failed to support either of the two hypotheses as the expected social class gradient in psychological disorder only emerged in the native group. Indian immigrants in fact showed a reversal with higher status individuals, especially women, having higher symptom levels. Overall Indian immigrants had significantly fewer symptoms than either of the other groups. The findings are discussed in terms of differential selection for migration and the varying experiences of immigrants after arrival in Britain.