This study investigated the role of strategy instruction on solution accuracy in children with and without serious math difficulties (MD) in problem solving. Children's posttest solution accuracy was compared on standardized and experimental measures as a function of strategy conditions. Strategy conditions included curriculum materials that gradually increased the number of irrelevant propositions within word problems. Children in Grade 3 (N = 193) were randomly assigned to one of five conditions: materials + verbal strategies (e.g., underlining the question), materials + verbal + visual strategies, materials + visual strategies (e.g., correctly placing numbers in diagrams), materials only-no overt strategies, and an untreated control. Compared to children with MD in the control condition, posttest outcomes for children with MD on standardized measures improved significantly under verbal + visual conditions, whereas posttest scores on the experimental problem-solving measures improved under the materials-only condition. Those strategy conditions found least effective made substantial demands on children's working memory capacity. The authors discuss benefits and limitations of strategy instruction.