• We conducted a study to determine whether performance on developmental tests at age 5 years could predict academic achievement at age 8 years. As part of a longitudinal developmental surveillance project, 179 children at risk due to perinatal complications who had passed developmental screening through age 2½ years and 50 comparison children underwent an extensive prekindergarten psychoeducational test battery at age 5 years and took the Iowa Tests of Basic Skills at age 8 years, if they had reached the third grade. The mean Iowa Tests of Basic Skills score was significantly lower for those children who were "flagged" on the prekindergarten psychoeducational test battery (t=5.39). Preacademic, rather than developmental, items appeared to be the best predictors. However, the prekindergarten psychoeducational test battery correctly predicted low achievement or grade retention in only 58% of cases. Its sensitivity was 0.45 and its specificity was 0.85. No significant difference was noted between group Iowa Tests of Basic Skills mean scores for the high-risk or comparison group. When low achievement and failure to reach the third grade were combined, prevalence of "failure" was higher for the high-risk group (31% vs 24%). The only perinatal variable predictive of low achievement was neonatal seizures. In summary, because the ability to predict future academic achievement at age 5 years is limited, routine developmental testing for symptom-free preschool children is not warranted. High-risk infant follow-up programs should focus on the first several years of life.