Our initial study compared 15 normally- developing and 13 language- delayed four- and- five- year- olds on a range of phoneme awareness tasks differing in the degree of explicit linguistic analysis required. The language- delayed group performed more poorly than the normally- developing children, and there were significant group differences on several tasks. A significant interaction effect reflected the particular difficulty the language- delayed group had with the more explicit tasks. Follow- up testing suggests that group differences are maintained over time and that the language- delayed children perform more poorly than the normally- developing children on tests of decoding and spelling at the end of first grade.An intervention study, training phoneme awareness skills in language-delayed kindergarten children, was undertaken with a new group of subjects. Fourteen language- delayed children participated in 16 training sessions over eight weeks. Fourteen normal and 14 language- delayed children served as controls. Only the language- delayed training group made significant gains from pre- to posttraining measures. Following training, the language-delayed training group performed similarly to normal controls and significantly better than language- delayed controls whom they had matched before intervention. One year later, the language-delayed children who received training maintained their gains on phoneme analysis tasks and performed significantly better than the language-delayed controls on reading measures. Educational implications are discussed.