Brewer and Smith (1984) showed that control mechanisms mediating speed-accuracy regulation contribute to retarded-nonretarded differences in processing speed, with poorly controlled trial-to-trial RT adjustments underlying the greater RT variability of retarded individuals. In Experiment 1, response deadlines controlled processing time, thus minimizing the influence of such control mechanisms. The obtained speed-accuracy relations showed that retarded subjects were unable to match nonretarded subjects' accuracy when responding as rapidly, thus indicating structural limitations on processing speed. The results of Experiment 2 showed, however, that significant adjustments to retarded subjects' processing speed-exceeding those produced by practice-are achievable. Extended training at a short deadline led to tighter control of RT adjustments, with substantial improvements in mean RT when subjects transferred to a self paced RT task.