The dual purpose of this study was (1) to design instruction in problem representation and solution of algebraic word problems in order to improve the algebraic problem solving skills of learning disabled (LD) adolescents; and (2) to devise and evaluate the effectiveness of instructional procedures for teaching LD adolescents to represent and solve algebra word problems. Twenty learning disabled students (Grades 8-10) were assigned to instructed (experimental) or control conditions. The experimental group received guided instruction in problem representation and problem solution for 600 algebra problems representing each of three mathematical structures: relational, proportion, and two-variable two-equation problems. Elements of the instructed phase included six pretests, intervention on alternate days for 12 weeks (using techniques which included scripts, verbal prompts, self-questioning, and reinforcement), six posttests, tests of near-transfer and far-transfer, and maintenance evaluation. The control groups received pretests, posttests, and their regular program of mathematics instruction. Examination of the single-subject data indicated that instructed students learned the three problem types, maintained what they had learned after a 6-week interval, and transferred their learning to problems deliberately constructed to be different. For two of three measures used, instructed students had significantly higher outcome scores than controls. (JW)