In developed countries, use of oral rehydration salts (ORS) solution with osmolarity lower than that of plasma has been recommended because of the risk of hypernatraemia. We compared the clinical efficacy of reduced-osmolarity ORS and standard ORS solutions in children with acute diarrhoea in four developing countries. 447 boys aged 1-24 months, admitted to hospitals in four countries with acute diarrhoea and signs of dehydration, were randomly assigned either standard ORS (311 mmol/L) or reduced-osmolarity ORS (224 mmol/L) solution. Total stool output was 39% greater (95% CI 11-75), total ORS intake 18% greater (3-33), and duration of diarrhoea 22% longer (2-45) in the standard ORS group than in the reduced-osmolarity ORS group. The risk of requiring intravenous infusion after completion of the initial oral rehydration was greater in children given standard ORS solution than in those given reduced-osmolarity ORS solution in three of the four countries (all-country relative risk 1.4 [0.9-2.4]). This relative risk was significantly increased only in non-breastfed children (2.0 [1.0-3.8], p < 0.05). In breastfed children, the relative risk of requiring intravenous infusion was not affected by the ORS solution (0.9 [0.4-2.0]). The mean sodium concentration 24 h after admission was significantly lower in the reduced-osmolarity ORS group than in the standard ORS group (135 [134-136] vs 138 [136-139] mmol/L, p < 0.01). Reduced-osmolarity ORS solution has beneficial effects on the clinical course of acute diarrhoea. Our findings support the use of reduced-osmolarity ORS solution in children with acute non-cholera diarrhoea in developing countries. Further studies are needed to find the best formulation and whether such a solution would be satisfactory for the treatment of cholera.