A validity study of the Self-Report Questionnaire (20 item version, Harding et al, 1980) against the criterion of a standardised psychiatric interview (Goldberg et al, 1970) was carried out at three primary care clinics in Sao Paulo; 875 patients filled in the SRQ-20 questionnaires and a subsample of 260 was interviewed by the psychiatrist. The SRQ-20 was shown to be a feasible screening instrument for psycho-emotional disturbance in these settings. Sensitivity was 83% and specificity 80%, and the questionnaire was a good indicator of morbidity. A correlation was found between questionnaire total scores and independent clinical judgment (r = +0.70). A sex difference in the validity coefficients is described.