The differential effects of 2 instructional strategies, an explicit schema-based strategy and a traditional basal strategy, on the acquisition, maintenance, and generalization of mathematical word problem solving were examined. Thirty-four elementary-aged students with mild disabilities or at risk for mathematics failure were randomly assigned to each of the 2 treatment conditions (schema and traditional). Results indicated that both groups' performance increased from the pretest to the posttest. All students were able to maintain their use of word problem-solving skills and generalized the strategy effects to novel word problems. However, the differences between groups on the posttest, delayed posttest, and generalization test were statistically significant, favoring the schema group. In addition, scores on the immediate posttest (77% correct) and delayed posttest (81% correct) for the schema group approached those of a normative sample of 3rd graders (M = 82% correct).