The longitudinal interrelations among age, visual and auditory short-term memory, and the concrete operational tasks of class inclusion and combinatorial reasoning were investigated over four testing occasions in a sample of 134 students, initially ranging from 5 to 12 years in age. The Piagetian task battery involved placement of colored triangles into superordinate, intersecting, and subclasses, and the systematic pairing of colored chips. Auditory sequential memory, memory span, and visual orientation memory tests involved digit series, word series, and two-dimensional geometric figures in varying spatial orientations. Year four logical operations performances were predicted from earlier variables by means of multiple regression analysis. Memory variables were significant predictors of later logical task performance in all regressions except in predicting final combinatorial reasoning status from year two data. Application of two models for the attribution of cause from correlational evidence indicated that combinatorial reasoning is a better predictor of short-term visual memory changes than the memory task is a predictor of combinatorial reasoning development. Visual memory should not, therefore, be regarded as a source of performance variance to be controlled in the investigation of logical operations development, but it should be seen as an outcome and a precondition for such cognitive restructuring. (Author/JAC)