Error correction is a ubiquitous instructional component for children with autism spectrum disorder (ASD). In the context of alternating treatment with repeated acquisition design, we taught four young Chinese children with ASD three sets of a match-to-sample task using multiple response repetition error correction with and without reinforcement. We assessed the participants' preferences of the procedures. Results showed that the acquisition rates were similar under both conditions. However, participant's preferences varied, with three participants preferring error correction with reinforcement and one preferring the without-reinforcement procedure. The discussion addresses the results from our comparison in light of prior studies and learner preferences of error-correction procedures, as well as the research and practical implications of our findings.