This study investigated whether in time-delay discrimination training, the performance of impulsive children can be improved by requiring self-monitoring of the correctness of nonwait responses, and to what extent these improvements are a function of the dimensions of the prompts. Four experiments, in each of which multiple prompts were used, one for each stimulus (S+, S-), were done. Comparisons between time delay of distinctive- and nondistinctive-feature prompts, with and without self-monitoring, were made across and within subjects. Time delay of distinctive-feature prompts without self-monitoring did not produce learning. The added requirement of self-monitoring nonwait responses led to a dramatic improvement in performance, but only when distinctive-feature cues were used for prompting and self-monitoring.