The use of saliva rather than blood for epidemiological studies of HIV prevalence, especially among injecting drug users, has several practical advantages. In a cross-sectional, behavioural and prevalence study among drug users in Glasgow during 1990, salivary samples were therefore obtained by the use of salivettes. Such samples were requested for anonymous anti-HIV testing from 498 persons in locations varying from residential rehabilitation centres to the open streets. Of this number, 35 refused to give a sample, resulting in a compliance rate of 93%. Of the 463 salivettes received by the laboratory, eight were found to be dry. Of the remaining 455 specimens, eight were found to be positive for HIV-1 antibody by means of an IgG antibody capture ELISA, so giving a prevalence rate of 1.8%. The results of testing saliva and blood spot samples collected at the same time on filter paper from 98 persons for HIV-1 antibody were 100% concordant. The study confirms the experience of others that specimens of saliva are easy to collect under variable conditions by non-medical staff and demonstrates that the salivette can provide an HIV antibody test result the same as that obtained from a blood spot. The prevalence of HIV antibody determined in this study is similar to that of other studies taking place in the city during the same period of time.