This study was designed to conceptualize and measure leadership behavior of principal. Specifically, an attempt was made to investigate the leadership styles of principals and develop Korean version of Leadership Behavior Description Questionnaire. Leadership behaviors were collected from 134 teachers who attend at the Graduate School of Education. They were asked to describe distinctive leadership behaviors of a variety of principals who have ever worked with them for many years of teaching. The first step in the selection of specific items from the large collection of 1200 was an attempt to classify them into the 8 areas which consisted of consideration, initiative, organization, production emphasis, directiveness, domination, representation, and fairness. The preliminary questionnaire was composed of 60 items which were selected among 200 items. This questionnaire was administered to 347 teachers. A factor analysis was made of intercorrelation of dimension scores. The principal component factor analysis revealed 9 factors with eigenvalue of 1.0 or higher. These accounted for 64 percent of the total variance. The following criteria were used to identify the basic factors: scree test, discontinuity of eigenvalues, interpretability, and the structure suggested by Halpin and Stogdill. Consequently, varimax rotations were carried out with 2, 3, or 4 factors. A three- factor solution appeared to be the most plausible for the present data and was finally' accepted. This solution yielded 50.1 percent of the total variance accounted for. Ten items each three factors survived respectively. They were arranged to form the Principal's Leadership Style Questionnaire(PLSQ). Three factors were named bureaucratic orientation, goal orientation, and teacher orientation. As estimates of reliability, the alpha coefficients for this sample were 0.91, 0.90, 0.92 for factors one to three respectively.