The acquisition of new skills may be hindered when teaching procedures vary from previously validated approaches or contain errors. In the present study, we compared the acquisition and maintenance of response chains taught using a perfectly implemented system of least prompts and a multiple verbal prompts procedure (i.e., addition of multiple verbal prompts and failure to follow through with more intrusive prompts). Four children, aged 6–9, participated in the study. An adapted alternating treatments design was used to compare the effectiveness and efficiency of learning during the system of least prompts and the alternative system of least prompts. Results were consistent with those obtained in previous studies in that the perfectly implemented and alternative prompting procedures were effective in teaching new skills for all participants. However, the perfectly implemented treatment required fewer trials to mastery for 4 of the 5 evaluations. Response chains taught under the multiple verbal prompts condition had poorer maintenance for 2 of the 5 evaluations. The results of the current study suggest that deviations from empirically identified teaching procedures may reduce the speed with which new skills are acquired.