This study examines socioeconomic conditions, psychosocial stress, and health among 264 infants, children, adolescents, and young adults aged 2 months to 18 years residing in a rural Caribbean village. Fieldwork was conducted over a 9 year period (1988-1996). Research methods and techniques include salivary cortisol radioimmunoassay (N = 22,438), systematic behavioral observations, psychological questionnaires, health evaluations, medical records, informal interviews, and participant observation. Analyses of data indicate complex relations among socioeconomic conditions, stress, and health. Household income, land ownership, parental education, and other socioeconomic measures are weakly associated with child illness. There is no evidence that apparent material benefits of high socioeconomic status--such as improved housing, diet, work loads, and access to private healthcare--have important direct effects on child health in this population. However, social relationships, especially family environment, may have important effects on childhood psychosocial stress and illness. Abnormal glucocorticoid response profiles, diminished immunity, and frequent illness are associated with unstable mating relationships for parents/caretakers and household composition. We suggest that family relationships and concomitant stress and immunosuppression are important intermediary links between socioeconomic conditions and child health.