The quality of data collected by surveys of sexual behaviour may be subject to sources of error involving reliability, validity, and non-participation bias. A preliminary study of sexual behaviour, including perception of risk for HIV infection, knowledge of HIV transmission and attitudes towards AIDS, was conducted in a genitourinary medicine clinic. The sample comprised 767 patients attending clinic over a 3-month period; comparisons were made between different methods of data collection (questionnaires and interviews) and between responders and non-responders. High levels of agreement were found between questionnaire and interview. Some sexual behaviours were more frequently reported at interview. No significant differences were found between responders and non-responders to either questionnaire or interview.