To date, very few studies have demonstrated the benefit of the differential outcomes procedure in human learning. In one of these studies (Maki, Overmier, Delos, and Gutman, 1995) there was evidence that normal children, ranging in age from 4 years and 6 months to 5 years and 5 months, performed better on a symbolic matching-to-sample task when they received differential outcomes following their correct responses. However, they only found facilitative effects of the differential outcomes methodology in 4-yearold children when the last eight trials were analyzed. In the present study, we used a similar task to that used by Maki et al. (1995) but we included a new phase for additional training. Participants, children ranging in age from 4 years to 4 years and 6 months, showed a better terminal accuracy and a faster learning of the task across the different phases when differential outcomes were arranged. These data indicated that additional training is not necessary to find the differential outcomes effect in 4-year-old children.