The effect of the self-monitoring training on reading accuracy and fluency of second-grade poor readers was examined. The participants were assigned in one experimental and three control groups. The experimental group was reinforced with token-economy for self-correction and fluent reading. One control group practiced reading without feedback, the second one practiced the calculation of simple arithmetic tasks, and the third one received no training at all. The improvement in reading accuracy was obtained only in the self-monitoring group, and remained for five months. However, the improvement in reading fluency was obtained in the two reading groups, but only the self-monitoring group showed further improvement during five months after training. The results confirmed the importance of self-monitoring already at the level of lexical access in reading and that of local text comprehension. (Contains 1 table and 2 figures.)