In this study we examine the pathways to schistosomiasis exposure and infection among residents residing on two islands (large, Qingshan; small, Niangashan) in the Dongting Lake region (Hunan province) of China. An exposure model, based on activity diaries, was used to quantify an individual's square-metre-minute (sq.m.min) daily water contact. Subjects living on the small island had a significantly higher (P=0.0002) degree of exposure (mean±S.D., 13.2±11.0 sq.m.min) than individuals dwelling on the large island (mean±S.D., 5.5±7.1 sq.m.min). Participants identified as stool egg positive (mean±S.D., 8.3±10.4 sq.m.min) had higher exposures than for those never treated (mean±S.D., 2.2±3.4 sq.m.min) for schistosomiasis, and these high exposures rose steadily to peak at 35-49 years of age and decline after age 50. This exposure pattern differs markedly from those reported for African or South American schistosomiasis. The majority of human water contact occurs on the lake. Egg-positive subjects reported significantly higher (P<0.05) episodes of water contact on the lake versus their egg-negative counterparts, who reported significantly higher (P<0.01) exposure at the aquaculture ponds. The results of path analysis revealed that sex, age, island of residence and whether a fisherman or not were the most highly significant independent predictors of lake exposure. This accounted for approximately 40% (R2=0.39) of the total lake exposure. Exposure to lake water was a strong predictor (P=0.0006) of past infection and a modest predictor (P=0.05) of current infection. Copyright (C) 1998 Elsevier Science B.V.