Recent reviews of research on individual differences in cognitive status during infancy have led to the conclusion that early test performances do not predict later IQ in any clinically useful way. These conclusions seem to be true for both normal children and for children born at developmental risk. In order to test an alternative strategy in prediction of later cognitive status from infant behavioral performance, 76 preterm infants were assessed with respect to differences in sustained attention when they were 8 months old. Scores on this measure were entered into a performance on 4 test scores at 2 years of age. The measure of sustained attention proved to contribute significantly to the prediction of later status on the Bayley Mental Scale and on the Gesell schedules. These sustained-attention scores were not significantly predictive of the scores on a Piagetian-based cognitive test, or on a assessment of receptive language. Subsequent analyses of these data for each gender separately revealed that increments in prediction were found for males only. The results are discussed in terms of current information-processing approaches to the study of stability of individual differences in cognitive functioning.