How did the study come about? The last decades of the 20th century witnessed a major epidemiological and nutritional transition in Latin America. 1–3 Mortality due to infectious diseases and malnutrition declined in most countries and neonatal deaths now account for a growing proportion of infant deaths. 4 , 5 In Brazil, infant mortality fell from 82.8 per 1000 live births in 19806 to 27.4 per 1000 in 2000. 7 There were also marked improvements in undernutrition: 37.1% of Brazilian children 98% of all deliveries in the city). All five city hospitals were visited on a daily basis by research team members. Eligible mothers included those living in the urban area of Pelotas according to the 1982 boundaries (part of the original city later became a separate municipality). They were interviewed soon after delivery (perinatal study) using a standardized, pre-coded questionnaire that was compatible with those used in the previous cohorts, but which also included a considerable amount of new information. How often have they been followed up? Differently from the previous two cohorts, the 2004 cohort included follow-up visits to the full cohort, rather than sub-samples. So far, visits have been carried out at the ages of 3, 12, 24 and 48 months. Learning from shortcoming of the previous cohorts, very detailed information on how to locate the family was collected since the first contact in the hospital. In addition to the family’s address and telephone number, information was collected on workplaces and on the addresses of relatives. The popularization of cell phones made it considerably easier to schedule the follow up visits. The 12-, 24- and 48-month follow-up rates (defined as the number of subjects traced plus the number of deaths in the period, divided by the total number of live births) were 94.0, 93.5 and 92.0%. A new follow up at the age of 6–7 years is scheduled for 2010–11. What has been measured? In the perinatal study, mothers were interviewed with regard to socio-economic, demographic and reproductive characteristics and on health care utilization, breastfeeding practices, lifestyles and morbidity. The interviews and newborn examinations were carried out in the maternity hospitals. The questionnaires, instruments and measurement techniques were highly compatible with those used in the 1982 and 1993 cohorts. Table 1 shows the groups of variables collected and measurements undertaken at birth (perinatal study) and in each follow-up visit. Newborns were weighed by the hospital staff using digital paediatric scales with 10 g precision, calibrated weekly to standard weights. The research team also measured birth length (using a locally made infantometer), head and chest circumference (using inextensible measuring tape). Gestational age was evaluated through three methods: the date of the last menstrual period, by ultrasound (when available) and by Dubowitz’s method. 12 Table 1 Variables collected and measurements undertaken at baseline and follow-ups, 2004 Pelotas Birth Cohort (2004–08) Follow-up at 3, 12, 24 and 48 months of age The mother or caretaker was interviewed and the child examined at home. Information was collected on socio-economic and demographic characteristics, health care utilization, feeding practices, lifestyles and child growth, development and morbidity since birth. Separate questionnaires on maternal health and child development were also administered. Perinatal and infant mortality study This sub-study was carried out in a partnership with the City Department of Health, and led to the creation of an audit system for perinatal and infant deaths. Deaths occurring among cohort children were recorded and causes of death ascertained by reviewing case notes of outpatient clinics and hospitals. Additionally, family members and the physician who looked after the child were interviewed. During the interviews a full history of the events preceding the death was obtained. As part of the death audit system, the reviewers also tried to identify potential failures in preventive and curative services that could have contributed to the children’s deaths. Feedbacks on the coverage of vital registration and on shortcomings in health care were provided to local policy makers. What is attrition like? The cohort profile in Figure 1 shows the number of children enrolled in the cohort and the number followed up at each visit. A total of 94 children had died and 3799 of the 4231 live birth children were visited at the 48 month follow-up. The proportions of children traced at the 48 month follow-up according to maternal socio-economic and demographic characteristics and by child sex and birth weight are shown in Table 2. Follow-up rates were highest (94%) among children born to older mothers (≥35 years) and lowest (∼90%) among those from families in the upper and lower extremes of income. Figure 1 Flow chart of the 2004 Pelotas Birth Cohort Table 2 Socio-demographic characteristics of mothers and children enrolled in the 2004 Pelotas Birth Cohort and percentage located at the last follow-up (48 months) Characteristics Original cohort (N = 4231) Percentage located a (%) P b Maternal age, years 0.02 25 kg/m 2 ) in 1982 and 34% in 2004. Birth intervals increased markedly in the period and there was an important reduction in the proportion of women who smoked during pregnancy—from 36% in 1982 to 25% in 2004. Antenatal care coverage increased over time and, by 2004, 94% of all participants were reached at the age of 4 years. A second limitation is that the cohort started at delivery rather than at the beginning of pregnancy. We considered this possibility but because there are close to 100 different facilities (public or private) providing antenatal care in the city, it was not feasible to identify mothers soon after they became pregnant. An additional limitation is that—with the exception of a few measures, such as weight and length at birth, Dubowitz gestational age, weight and height at follow-ups, and child development tests—the vast majority of information was obtained through maternal reports. The quality of reported information will therefore depend on characteristics of the mother being interviewed (such as age and schooling), and also on the nature of the variable being collected (e.g. how sensitive the information may be, and what is the duration of recall). Can I get hold of the data? Where can I find more? Joint analyses of the cohort data are welcome and we have collaborated successfully with investigators from the UK (University of Bristol) in these early analyses. The most fruitful collaborative experiences from our older cohorts including having doctoral or post-doctoral fellows from other institutions spend a period of time in Pelotas, or for our students or fellows to spend time in other institutions, thus helping build local capacity. For interested young researchers from Latin America, the Wellcome Trust sponsored post-graduate programme in Life Course Epidemiology was launched in 2005 and has so far trained 15 MSc and PhD students from Latin America and Africa. They receive full scholarships to work on the cohorts. Applications are welcome. For further information contact the program website at http://www.epidemio-ufpel.org.br/projetos_de_pesquisas/estudos/ coorte_2004 or e-mail the corresponding author. Funding The 2004 birth cohort study is currently supported by the Wellcome Trust Initiative entitled Major Awards for Latin America on Health Consequences of Population Change. Previous phases of the study were supported by the World Health Organization, National Support Program for Centers of Excellence (PRONEX), the Brazilian National Research Council (CNPq), the Brazilian Ministry of Health, and the Children’s Mission. Conflict of interest: None declared.