To assess the impact on child growth of the nutrition-counseling component of the Integrated Management of Childhood Illnesses (IMCI) strategy, a randomized trial was implemented. All 28 government health centers in a Southern Brazil city were paired according to baseline nutritional indicators. One center from each pair was randomly selected and its doctors received 20-h training in nutrition counseling. Thirty-three doctors were included and 12-13 patients < 18 mo of age from each doctor were recruited. The study included testing the knowledge of doctors, observing consultations and visiting the children at home 8, 45 and 180 d after the initial consultation. Maternal knowledge, practices and adherence to nutritional recommendations were assessed, and anthropometric measurements were taken. Day-long dietary intake was evaluated on a subsample of children. Doctors in the intervention group had better knowledge of child nutrition and improved assessment and counseling practices. Maternal recall of recommendations was higher in the intervention than in the control group, as was satisfaction with the consultation. Reported use of recommended foods was also increased. Daily fat intake was higher in the intervention than in the control group; mean daily intakes of energy and zinc also tended to improve. Children 12 mo of age or older had improved weight gain and a positive but nonsignificant improvement in length. Nutrition-counseling training improved doctors' performances, maternal practices and the diets and weight gain of children. The randomized design with blind outcome evaluation strongly supports a causal link. These results should be replicated in other settings.