In laboratory symbolic matching-to-sample (MTS) tasks, acquisition commonly proceeds via trial and error, whereas in applied settings, MTS instruction typically includes prompting and prompt-fading strategies. We examined the effects of error-contingent prompts in symbolic MTS under sample-first and comparison-first presentation arrangements. Three children (4-6 years) participated in each of two experiments. Each participant received two instructional sessions in each of four conditions, with new stimuli in each session. Dependent measures included terminal accuracy, items mastered, and overall efficiency across participants. In Experiment 1, error-contingent prompts facilitated acquisition under the comparison-first but not under the sample-first arrangement. In Experiment 2, the temporal arrangement of error correction trials was altered such that prompts did not delay the presentation of comparison stimuli. With this change, error-contingent prompts facilitated acquisition under both stimulus presentation arrangements. The results suggest that in a sample-first presentation format, prompts may be ineffective if they delay comparison presentation.