The aim of the study was to compare the development of children with and without a programme of psychosocial stimulation in 'control' and 'intervention' sites in a poorly resourced area of northeast Brazil. The sample (n = 156, born 1998) was from a larger cohort. The cohort was tested at 12 months (baseline) with the Bayley Scales of Infant Development. All children in the intervention site with a mental development index (MDI) and/or psychomotor development index (PDI) < or = 100 were enrolled. Each time such a child was enrolled, the next child tested in that site of the same sex and with an index of 101 to 115 was also enrolled, and the next two children matched for sex and scores of < or = 100 and 101 to 115 in the control sites were recruited in parallel. The intervention comprised 14 contacts between 13 and 17 months of age. All children were tested again at age 18 months. The intervention and control groups were similar at baseline for a range of socioeconomic, demographic, environmental, and biological variables, and their MDI and PDI were also similar. At 18 months, the mean differences between the intervention and control groups were + 9.4 points for MDI and + 8.2 points for PDI (p < 0.001 in each case). For children with an initial score of < or = 100, the mean difference between the intervened and control groups was + 11.2 points for MDI (p < 0.001), and + 10.8 points for PDI (p = 0.001). The intervention was thus associated with significant improvements in cognitive and motor development.