This study evaluated the effectiveness of phonological awareness instruction with 61 kindergarten children in two schools who had been identified as low in phonological awareness. The children received either: (1) instruction at the phoneme level only, (2) instruction at the onset-rime level before instruction at the phoneme level, or (3) no intervention. Fourteen additional participants in a non-equivalent condition received a subset of tests. The interventions were delivered in 15-minute sessions, four times per week for nine weeks, in groups of three to four children. Group results were compared with pre/post and slope analyses. Children in the experimental groups performed reliably better on phonemic segmentation fluency, onset recognition fluency, and blending at posttest than children in control groups. No reliable differences between groups were found on phonological awareness measures that required generalization or on rapid retrieval on alphabetic understanding measures. Instruction at the phoneme and onset-rime levels were equally effective and efficient. (Contains 92 references.) (DB)