Aims: To assess the reliability of saliva samples as a means of testing for HIV-antibodies outside clinic settings. Methods: Men taking part in a non-clinic longitudinal study of homosexually active men provided samples of saliva and blood. Sera was screened using a competitive ELISA (Wellcozyme) and positive sera were confirmed by an indirect ELISA (Abbott). Saliva samples were screened either using an IgG captive radioimmunoassay or an amplified ELISA. Results: A total of 534 paired saliva and blood sample were tested. Overall sensitivity was 96.2% and specificity was 100%. None of the saliva tests were falsely positive for HIV-1 antibodies. Conclusions: HIV-1 saliva tests can reliably be used in a non-clinic or field setting. However, if results are to be given to respondents, it is necessary to offer adequate counselling and consider the mechanisms for referral and follow-up for those that are found to be HIV-1 antibody positive.