This paper reports on a study of the effects of problem-solving computer software on the problem-solving ability of secondary school mathematics students. A cognitive science-based definition, instrumentation, and instructional pedagogy were used. Members of an experimental group were involved in computer-augmented mathematics instruction, while members of a control group were involved in more traditional mathematics instruction. Although results of a statistical analysis of a test of problem-solving ability were mixed, the experimental group demonstrated Jignificantly better performance on standardized tests of mathematics content. Implications of results for mathematics instruction are discussed. (Keywords: mathematics, problem solving, achievement.) "Problem solving (means) the behaviors that researchers who say they are studying problem solving, study." This concise statement by Bernard Corman ( 1957) suggested the ambiguity that surrounds much of the work done in the investigation of mathematical problem solving.Anecdotal comments by mathematics teachers often suggest the need for clarity in discussions of problem solving. Additional anecdotal comments by the same teachers also suggest the need for new methods of instructing in problem solving and of increasing the motivation of their students. Similarly, many educational reports have stressed the need for increased attention to instruction in problem solving. Much of the current literature in mathematics education recommends new methods for developing problem-solving skills.This article and the related research grew out of efforts to work with a group of high school mathematics teachers to clearly define the concept of mathematical problem-solving and to develop procedures to enhance the problem-solving ability and motivation of their students. The definition was based in cognitive science. The procedures centered around a recent technological development: problem-solving software for a microcomputer.