The present study aimed at investigating developmental changes in cognitive strategy for concept learning.Ss were required to identity the criteria of concepts and to classify a set of stimuli, utilizing information transmitted by positive or negative instances. A modified and objectively defined procedure of Vygotsky-type concept formation experiment was undertaken. Variable dimensions of 22 stimuli were size (2values), form (6), color (5) and the presence of a frame (2).These22stimuli could be classified into 4 categories according to their values on relevant dimension, size and frame. Ninty-five Ss, who were5-6, 7-8, 9-10, 11-12, 13-15, 20-above-years-old, participated in the experiment. Their cognitive strategy, i.e., the mode of information seeking and utilization was diagnosed on the basis of structural aspects of1st and 2nd classification and the level of verbalization of concept intentions. As results, 3 developmental stages were identified. Various behavioral indices of Ss which belonged to each stage were compared in detail. Ss who belonged to the Stage I, who were mostly kindergarten children, could not classify a set of stimuli according to any discernible principle. They seemed not to comprehend the presence of a criterion. They could neither utilize information brought about by new “samples”, nor verbalize the concept intention even after all of the stimuli were placed into relevant categories. Ss who belonged to Stage II, mainly2-8graders, could identify the concept intention inductively and apply it to other stimuli.Sometimes, their first few calssifications had no discernible criterion, but they could discover and verbalize relevant dimensions after a few instances were presented. Suggestion for drawing their attention to relevent dimensions were effective in facilitating their performance. Ss at Stage III (above the6-graders) classified stimuli according to their hypothetical, deductivelyintroduced criterion from the start.When they were informed their former classification was not adaptive, they tried another hypothesis until relevant categories were identified.