The predictive validity and the general usability of a battery of 10 tests reported by de Hirsch, Jansky, and Langford, the de Hirsch Predictive Index Tests of reading failure, were examined. The de Hirsch battery was administered to 433 kindergarten children in six public schools. When the pupils entered first grade, the Metropolitan Readiness Test was administered; in second grade, pupils took the Gates-MacGinitie Reading Test. Regression analysis was used to assess the degree of additional information the de Hirsch tests added to the prediction of scores on the Gates-MacGinitie, beyond that provided by the Metropolitan. The alphabet subtest of the Metropolitan and the word reversals and word reproduction tests of the de Hirsch contributed the most to the prediction of second-grade comprehension scores. Discriminate analysis showed that the Metropolitan Test used alone properly classified 28 of 29 poor readers and 68 of 79 superior readers, and that the alphabet subtest discriminated nearly as well as all six Metropolitan subtests. It was concluded that it is questionable whether the time and expense involved in administering the de Hirsch tests is justified for general testing, but that they may be useful for assessment of marginal kindergarten students. Tables and references are included. (VJ)