Two studies carried out in 1982 and 1993 in the city of Pelotas, Southern Brazil, provide a unique opportunity for assessing the impact on maternal and child health of the economic and health care changes, which took place in Brazil in this period. The cohorts of mothers and infants of 1982 and 1993 were studied from the time of delivery. In both years, all mothers identified in the city's maternity hospitals answered a standardised questionnaire and their infants were examined. Over 99% of all children born in the city in each of the 2 years were included in the cohorts. Deaths occurring among these children were monitored prospectively, as well as all hospital admissions in the 1993 cohort. In the 1982 study, attempts were made to locate a 25% sample of the children at the mean age of 12 months using the addresses collected at the hospital (82% of the children were located), and all of the cohort children at the mean age of 20 months and 42 months, through a city census (87% were located in both follow-ups). In the 1993 study, 20% of all children plus all low birthweight infants were sought at 12 months of age, using the addresses collected at the hospital, and 95% were successfully traced. There was a 12% fall in the number of births occurring in 1993 (5,304 births), in comparison with 1982 (6,011 births), in spite of the increase in the population of reproductive age in the city during the decade. There was a marked difference in maternal height and weight at the beginning of pregnancy, with women giving birth in 1993 being, on average, 3.4 cm taller and 2.5 kg heavier than those who gave birth in 1982. The proportion of preterm babies (<37 weeks), measured by the date of last menstrual period, increased from 5.6% in 1982 to 7.5% in 1993. The median duration of breast feeding increased from 3.1 months in 1982 to 4.0 months in 1993. At 12 months of age, the prevalence of deficit of weight for age decreased from 5.4% in 1982 to 3.7% in 1993. The prevalence of deficit of height for age, however, increased from 5.3% to 6.1%. The perinatal mortality rate dropped 31%, from 32.2 per 1,000 births in 1982 to 22.1 deaths per 1,000 births in 1993. There was also a marked reduction in the infant mortality rate, from 36.4 per 1,000 livebirths in 1982 to 21.1 per 1,000 livebirths in 1993. The findings of the study indicate that there were improvements in the decade for most of the indicators evaluated, with the exception of birthweight and gestational age. It appears that improvements in perinatal and infant mortality rates are largely due to improvements in the health care sector.