In 1983, a cohort study to follow up the family contacts of leprosy cases was implemented in French Polynesia to assess the usefulness and applicability of phenolic glycolipid-I (PGL-I) serology in a leprosy control program. A total of 1201 contacts (666 females, 535 males) have been included in the study. The IgM anti-PGL-I seroprevalence determined on the initial sera was 17%. It was significantly higher among females than males (20% vs 15%, p = 0.02). From 1983 to 1992, 4 out of 204 (2%) anti-PGL-I seropositive contacts developed the disease (1 indeterminate, 1 BT, 1 BL, 1 LL) compared with 10 out of 997 (1%) seronegative contacts (4 indeterminate, 3 BT, 1 BB, 2 TT). Of these 10 patients, only 3 (2 indeterminate, 1 BT) converted to seropositivity when leprosy was diagnosed. The risk of developing leprosy was not significantly higher among seropositive than among seronegative groups (2% vs 1%, p = 0.2). A PGL-I circulating antigen test performed on 216 selected sera at entry into the trial showed a higher antigen prevalence when the antibody level was higher. PGL-I antigen was detectable in 5 of 12 patients tested prior to diagnosis (1 LL, 1 BL, 3 indeterminate). The median time to externalize the disease was not significantly different among antibody-positive and -negative contacts (17 vs 25 months, p = 0.3). The relative risk of developing leprosy for contact individuals was 30.8 times that of noncontacts, and 15% of the total new cases detected between 1983 and 1992 emerged from the study population.(ABSTRACT TRUNCATED AT 250 WORDS)