The prereading and reading performance of a group of 370 predominantly Hispanic, bilingual public school children was followed for six years. The aim of the study was to compare the factorial make-up and predictive validity of tests used in the current research with findings from a previous study. The early study had followed the children from the end of their kindergarten year until the end of second grade. For purposes of the present study a 13-test predictor battery was administered to the children at the beginning of their first-grade year. This battery included a number of measures used in the early study, among them a 5-test Screening Index developed in the course of that study. Records were obtained of the follow-up children's 2nd, 3rd, 4th, 5th, and 6th grade reading test scores. Factor analysis of the pool of 13 predictor tests showed a two-cluster battery structure. These clusters closely resembled two of the four factors defined in the early study. As used with the group of bilingual children, the Screening Index misclassified many of the failing readers, as well as others who went ahead to read at grade level. The Screening Index quite accurately identified, across grades, children who eventually read well. A number of predictor tests administered at beginning of first grade continued to show significant correlations with reading at all grade levels. The consistent contribution of the predictor tests with reading points to their continuing usefulness as a point of departure for assessing children's readiness to read.