BACKGROUND Since 2000, there has been a decline in the proportion of oral rehydration salts (ORS) therapy in childhood diarrhea. How to sustain and achieve a high level of ORS therapy continues to be a challenge. METHODS The data of 14112 households and 894 villages in 45 counties across 10 provinces of Western China were collected in 2005. Generalized estimated equation logistic regression models were used to identify the determinants of ORS use in home-based and village-level care. RESULTS The therapy rate of ORS was 34.62%. This rate in home-based care (HBC) was significantly lower than that in village-level care (VLC), township-level care or county-level-or-above care. The children in the families with several pre-school-aged children (OR = 0.29 95% CI: 0.10, 0.86) or of the smaller age (12 vs 36 months: OR = 0.10 95% CI 0.02, 0.41; 24 vs 36 months: OR = 0.26 95% CI 0.09, 0.77) were less likely to receive ORS therapy against diarrhea in HBC. The children whose family had the habit of drinking boiled water (OR = 2.77 95% CI 1.30-5.91), or whose caretakers received educational materials about childhood diseases (OR = 3.08 95% CI 1.54, 6.16), or who were living in the villages in which village clinics had the available ORS packages (OR = 3.94 95% CI 2.25, 6.90) were more likely to receive ORS therapy against diarrhea in VLC. CONCLUSION There thus, ORS promoting program should give the highest priority to home care. ORS promoting strategies for low-level care could be strengthened based on children characteristics, the habit of drinking water and the situation of receiving educational material in the families and on the availability of ORS packages in village clinics in rural Western China.