A cohort of public school first-graders living in Embú, a large urban poor community located in São Paulo City, will be followed-up for two years in order to examine predictors of early grade retention. All state public schools located in Embú and connected with the Medical University (12) will be eliginle to enter the study. All first- grade classes (51) will be represented inthe initial sample to maximize interclassroom diferences. The sample will be stratified by classroom and students will be randomly selected within each class. Only students starting grade one for the first time will participate in the study to guarantee sample homogeneity in terms of previuous exposure to acadmic curriculum. The sample will be followed-up from the begining of grade one to the begining of grade three, and information will be gathered on child, family, teacher, classroom, and school characteristics at school entrance and/or during follow-up.

At the begining of grade one, children will be tested for readiness skills (IQ) and submitted to pediatric clinical examination. At the same ocasion, parents will be interviewed in order to report child behavior problems, child physical health problems, family stressful situations, parent education and family income.

During grades one and two, classroom observations will be conducted at random to collect information on classroom facotrs ( classroom climate, classroom time devoted to instruction, classroom teacher changes). Observers will be naive to the results obtained at school entrance, and interobserver and test-retest reliability will be assessed using video-tape observations. Furthermore, data will be collected on child absenteeism based on classroom observation and teacher report.

At the end of grades one and two, teachers, principals and parents wil be interviewed to give information on events occurring in the past year. Teachers will report on child behaviour problems in the classroom, and on teacher factors (satisfaction with job, previous experience in teaching, additional education or training, acceptance of student); principals will report on school factors ( school incentives for staff ongoing education, schol-community links); and parents will report on parental involvement in school. Also at the end of grades one and two, students will be submitted to independent achievement tests on reading and mathematics, and school records will be examined in order to register marks on final exams. Grade retention at the end of grade two will be registered according to school records, and confirmed ar the begining of the next school year by observing the presence of grade failures in second -grade classes.

Stepwise regression analysis will be applied to five sets of predictors, wich wil enter the regression equation in a pre-established hierarchical order. The first set includes child-family factors at school entrance; the second set includes child -family factors measured during grade onde; the third set includes child-family factors measured during grade two; and the fifth set involves teacher-classroom-school factors at grade two. Analysis results will show the percentage of early grade retention variance explained by each set of variables. In addition, statistically significant individual variables will be identified.