Two experiments were conducted to determine the elicited imitation performance of preschool children when the stimulus sentences were meaningfully related to each other in a story as compared to when the sentences were arranged in a random order. Experiment 1 used a repeated-measures design and found a practice effect demonstrating less errors during the second trial regardless of experimental condition. Experiment 2 implemented a group design with one group imitating the related (story) sentences and the other imitating the randomly arranged sentences. The results showed no significant difference in the mean number of errors occurring in the "story" versus the "random" condition. The implications of the practice effect found in Experiment 1 are discussed and the error pattern during the second trial is evaluated. The efficacy of using discursively related as opposed to unrelated sentences in elicited imitation tasks is discussed from a pragmatic point of view.