This article presents the methodology and baseline findings of a large multicenter trial involving four countries from Latin America (Argentina, Brazil, Cuba, and Mexico). The study was a randomized, controlled, single-masked trial to investigate the impact of social support during pregnancy on perinatal outcomes. Pregnant women with gestational ages between 15 and 22 weeks were screened in health facilities in the four countries. Those presenting with one or more risk factors for having a low-birthweight baby were invited to join the trial. A total of 2235 women--between 500 and 600 in each country--were randomized into an intervention (n = 1110) or a control (n = 1125) group. Both groups were comparable in terms of nearly all baseline variables. The intervention group received a minimum of four visits at home by a trained health worker who provided direct emotional support, health education, and an attempt to enhance the woman's social support network. Over 90% of all women were evaluated at 36 weeks of pregnancy and soon after delivery, and 85% at the 40th day postpartum. The outcomes under study included intrauterine growth retardation, gestational age, perinatal and maternal morbidity and mortality, labor interventions, psychological distress and characteristics of the social support network, among others. This trial showed that it was possible to select, screen, randomize, visit, and evaluate a large number of women in four Latin American countries using a standardized methodology.