The double-classification problem forms the basis of many intelligence test items and is of major interest in Piagetian theory. The solution of such a problem requires the child to take into account simultaneously two different dimensions of stimulus variation (e.g., size and shape) while inferring a logical relation. The present study utilized highly structured yet individualized training techniques in an attempt to teach the skill of solving double-classification problems to first-grade children. Following a pretest of double-classification skill with stimuli varying in color and shape, 42 first-grade Ss matched for pretest score were randomly assigned to either an experimental or a control group. Experimental Ss were individually instructed by E in the skill of solving double-classification problems with stimuli that varied in color and shape. Instruction was terminated when S reached a predetermined level of proficiency, or when a half hour elapsed. Control Ss were given individual attention for an equivalent amount of time while engaged in an irrelevant task. Several hours following the treatment, each S was retested by a second E who had no knowledge of what group S was in. Both a posttest (a randomly parallel alternate form of the pretest involving the dimensions of color and shape) and a transfer test, involving the dimensions of shading and size, were administered. Experimental Ss scored significantly higher than control Ss on both learning and transfer tests. Ss were tested for retention four months later. Again experimental Ss scored significantly higher than control Ss on both learning and transfer tests. The results are discussed in terms of Piagetian theory, and implications for intelligence testing are considered.