Objective To implement and evaluate a national survey of sexual behaviour using computer-assisted telephone interviewing (CATI). Design A two-stage stratified national sample survey in which households were selected by random digit-dialling (RDD), with a single eligible interviewee per selected household, followed by subsample surveys of non-contacts and refusals to determine eligibility. Methods A 15-minute questionnaire based on the Global Programme on AIDS (GPA)/World Health Organization (WHO) protocol was administered by telephone to a nationally representative sample of 2361 respondents in the 18–54-year age group. Results The overall response rate was 63%, but lower in the cities, in the 18–24 age group, and among men. Three-quarters of surveyed non-contacts, and a quarter of resurveyed refusals, did not meet the eligibility criteria for the study. Less than 20% of refusals cited the subject matter of the survey as the reason for refusal. Item non-response (< 1%) increased with question sensitivity, and varied by respondent age, ethnicity and partnership status. Men reported twice as many adult lifetime partners as women. Conclusions The GPA/WHO protocol can be successfully adapted to administration by telephone, with adequate response rates and exceptionally low levels of item non-response. CATI is a cost-effective method for collecting national information on sexual behaviour in countries where there is a high level of telephone ownership. Used in conjunction with RDD, it can overcome problems of sample design in settings where there is no comprehensive population-sampling frame. Checks on item sensitivity and partner estimates suggest that acceptable levels of reliability can also be achieved.