An epidemiological method, field-tested in Hunan, China, to identify residents potentially susceptible or insusceptible to endemic schistosomiasis japonica is described, as a prelude to selection of subjects for immunogenetic studies. After an initial cross-sectional survey on two islands (Qingshan and Niangashan--population 2990) in 1995-1996, an informative cohort (N = 249) was selected for treatment and 9-month follow-up to measure exposure and re-infection. Both the population prevalence (15.8%) and the geometric mean intensity of infection (26.2 eggs/g faces) indicated that the islands were moderately endemic for schistosomiasis. Exposure measurements revealed a strong, positive, linear association (r = 0.70) between daily activity diaries and direct water-contact observation. Individuals identified as stool-positive for schistosomiasis had significantly more water contact than those who were egg-negative (P = 0.03). Almost all (93%) of the cohort had ultrasonographic evidence of periportal fibrosis before treatment but in only 1.2% was this fibrosis scored > 1 in terms of the stages identified by the World Health Organization. At the follow-up it was possible to classify the 249 subjects into three, distinct, exposure-infection epidemiological groups. The first group (N = 20) was susceptible to re-infection and constituted 8% of the cohort. The second group (N = 61) was apparently insusceptible to re-infection despite the continuing high levels of exposure and included 24% of the cohort. The other 68% of the cohort (N = 168) remained uninfected but were at most only moderately exposed, or had a status indeterminate due to non-compliance. This epidemiological identification of susceptibles and insusceptibles for schistosomiasis japonica' links field and ongoing laboratory studies aimed at characterising the genetic and immunological factors associated with resistance to re-infection and/or disease.