The authors investigated the effects of student spatial ability, as measured by Raven's Progressive Colored Matrices (J. C. Raven, 1938) and type of instructional program on geometry achievement. Sixth-grade students worked through either 6 instructional activities in Geometer's Sketchpad (Key Curriculum Press, 1993), a dynamic geometry program, or a geometry tutorial, both of which paralleled Connecticut's geometry standards. The authors hypothesized that students working with the activities in Sketchpad would learn geometry better and that (after controlling for mathematics ability) student spatial ability would predict success in such an environment more reliably than it would in the tutorial program. However, students with high spatial ability performed significantly better than did low-spatial learners in both instructional treatments; students in the Sketchpad treatment scored only marginally higher on the posttest than did learners in the tutorial condition, despite spending more time on task. Results have implications for mathematics instructors, researchers, and computer-based programs.