A eurvey on friendship structure of an elementaly school class was continued extending over three years. Monthly, every child in the class was interviewed individually by the home-room teacher and asked whom of his classmates he liked or disliked. The first interview was performed a week after the entrance. choice-data thus gained were analysed quantitatively in the main, and the following findings were derived of the interpersonal relations in the class. (1) Being less than unit in early stage, mean number of classmates children mentioned as "Likes" increased gradually and overcounted three in final stage. Mean number of "Dislikes," a little less than that of "Likes," followed a similar course. (2) In positive choice, girls selected more girls than boys almost from the beginning, while boys selected significantly more boys than girls not before than the third term of the second grade. In negative choice, on the contrary, both sexes selected much more boys than girls from the beginning on. (3) The star-status (in the sense of receiving 2.S.D. more choices than the average) was occupied almost exclusively by girls in positive choice and by boys in negative. Their positions, especially of positive one, were considerably stable. (4) In the age-range which the present study covered no differences due to the developmental stages were found in the consistency of choice behavior. On the other hand, it was found more consistent in positive choice than negative choice at every stage. (5) The existing cliques, defined as two or more individuals all of whom choose each other mutuallywere more frequently found among girls than boys. Though the intra-sexual cliques of both sexes tended to increase, the inter-sexual cliques were less frequent in the third grade than in the second. As conclusions from the above findings it may be said that the sexual cleavage appeared at earlier stage than generally stated. And that in was initiated by girls-side, boys following with a delay.