Rates of sexual-partner change and patterns of high-risk behaviour are important determinants of the spread of HIV. We carried out a survey to assess the feasibility of studying sexual lifestyle in a random sample of the British population, aged 16-64 years, in November 1987. Two thousand and seventy-seven households were selected using a multi-stage probability sampling procedure. Seven hundred and eight-five adults participated in a structured interview. The schedule included demographic details, attitudes to AIDS, numbers of sexual partners in different time periods, history of homosexuality and contact with prostitutes. An interview was obtained in 61% of households where contact was made, but the overall response rate was low (48%). There was marked variability between individuals in numbers of sexual partners in given time intervals. Men and women in younger cohorts had experienced first sexual intercourse earlier and had higher numbers of sexual partners than people in older cohorts. Surprisingly few reported high-risk behaviour such as homosexuality and use of prostitution. The methodological problems in trying to obtain unbiased and valid data on sexual behaviour are discussed. Further work is necessary to improve the response rate and questionnaire design.