Five studies assessing the predictive validity of the Denver Developmental Screening Test (DDST) were evaluated and the results were pooled. Only one study demonstrated good compliance with the authors' methodological standards, but the results were similar to the over-all pooled results. The pooled data indicated that for children over three years of age: (1) the DDST is specific (94 per cent of the children with a good outcome were categorized as normal); (2) the DDST is not sensitive (it did not identify 80 per cent of the children who later had a poor outcome); and (3) a child with a poor outcome was 14 times more likely to have an abnormal or questionable DDST result than a normal one. These data suggest that a child with an abnormal DDST is likely to have a poor school outcome and that many children with school-related problems who might benefit from early intervention are not identified by the DDST.