The China Health Insurance Experiment (CHIE) was designed to test insurance as an alternative for health care financing in rural China. Six experimental groups were assigned to different insurance plans in 1989; each group switched to a new plan in 1990. The insurance plans differed according to coinsurance rates. In this study, data from 7,333 panel sample enrollees were used to estimate outpatient and inpatient demand functions for episodes of care. Demand functions were estimated using negative binomial regression (NBR); coinsurance elasticities were also estimated using first difference (fixed effects) models to reduce bias from unobserved heterogeneity. The principal findings are: (1) in rural China coinsurance exerts a significant and negative effect on demand; (2) coinsurance effects are similar for different population sub-groups and for different types of episodes; (3) elasticities of demand varied according to access to care conditions; (4) patterns of socio-demographic effects were similar to those reported on North American populations. These results provide support for proponents of health insurance as a primary means of financing rural health services in low-income developing countries.