In physical education courses, ability to learn sports and/or movement has to be developed, along with movement skill and physical fitness. This competence would depend greatly upon attitudes toward the physical education coursed. Most of the attitude scales, as well as Wear Attitude Inventory, have been inductively developed by the goals of physical education. It is, however, considered that attitude measurment should be dependent on the structure of attitudes among children toward physical education courses. The present study was designed to develop attitude scales for the 3rd and 4th grade children in elementary school using factor analysis, and to confirm its validity as a mean of analyzing the effect of the instruction. The detailed description of the method has been given by Umeno (1980). The results obtained were as follows: 1) The structure of attitudes among the 3rd and 4th grade children in elementary school toward physical education courses could be explained by the following four factars: The first factor could be named "Evaluation" for the instruction of physical education, and the second could be named "Value" for the physical education. Than the third could be named "Pleasure" for physical activities. In addition to the above factors, "Value of Sports Life" was extracted from the second factor in the 4th grade children. 2) In an attempt to construct attitude scales originated in the structure of attitudes among children toward physical education courses, items with the value of more than Mean + S.D. of the absolute values of factor loading were selected. 3) Significant coefficients of Split-half Reliablity were found in "Evaluation" and "Pleasure" scales for 3rd grade children, and in "Evaluation", "Pleasure" and "Value" scales for 4th grade children. 4) In terms of the attitude scales which exhibited higher reliablities, significant negative correlations were observed between attitude scorers at the beginning of a term and the magnitudes of change in the score during the term. 5) Fromthe above results 3) and 4), it was suggested that "Pleasure" and "Evaluation" scales in the 3rd and 4th grade children and "Value" scale in the 4th grade children had the validity as a mean of analyzing the instruction, respectively.