Following pretesting with Raven's Coloured Progressive Matrices Test, 48 Ss were randomly assigned to either experimental (remedial self- instructional loops for missed pretest items) or control (no instruction) conditions. A posttest revealed that the remedial loops were successful, but that their effects did not transfer to other items. A retention test given two months later showed no difference between groups. It was concluded that self-instructional procedures could be used to increase children's scores on the Coloured Progressive Matrices Test. Further tryout and revision of procedures will aim at making the increase more general and lasting.