The purposes of this study were to determine (1) whether 3-, 4-, and 5-year-old children can segment words, and (2) whether they can profit from instruction in segmentation. In the first phase of the study, all subjects (N=68) received a 28 word test designed to assess children's performance on four different segmentation tasks. Since the children could not read, test items were illustrated and presented orally with appropriate distractors. In the second phase, children who did not attain the criterion of 80% correct on the pre-test, received instruction in segmentation tasks and a post-test. ANOVAs and post hoc procedures yielded the following results: significant differences were found in the performance of 3-, 4- and 5-year-olds; significant differences were found among the four segmentation tasks; subjects learned from instruction in segmentation; certain tasks were easier than others; and disyllable tasks appeared somewhat easier than monosyllable tasks. (Author/MP)