In a randomized, double-blind vaccine trial in Venezuela, about 29,000 contacts of leprosy patients have been vaccinated with either a mixture of heat-killed Mycobacterium leprae and BCG or BCG alone, and are being re-surveyed annually to detect new cases of leprosy. All contacts had a serum sample collected at the time of entry into the trial, and 13,020 of these sera have been analyzed for antibodies to phenolic glycolipid-I (PGL-I). Antibody levels have been related to various characteristics of the contacts and to their risk of developing leprosy in the following 4 years. A strong association was found between PGL-I antibody level and the risk of developing leprosy, in spite of possible modification of the incidence rate induced by vaccination. Antibody levels were higher in females than in males, and declined progressively with age. Household contacts had higher levels than did non-household contacts, and levels were higher in individuals from the state in Venezuela which has the highest incidence of the disease. No substantial differences were found in antibody levels between contacts of multibacillary and paucibacillary patients, which may in part reflect the influence of treatment, and there was no clear association with the presence of BCG or lepromin scars or with skin-test responses to PPD and leprosy soluble antigen. The assay of antibodies to PGL-I seems unlikely to provide a sensitive or specific test for infection with M. leprae, and measuring PGL-I antibody levels as a screening procedure to identify those at high risk of developing leprosy is unlikely to be particularly useful in most leprosy control programs. Such assays may be useful for the epidemiological monitoring of changes in the intensity of infection with M. leprae in a community and for the study of carefully defined groups of contacts during some phases of control programs.