This stndy is the first step in investigating what effects “moral” classes have on the character development of children and is aimed at the analysis of the teacher-pupil interaction during “moral” education. For this purpose three records of instructions by different teachers each pursuing one and the same aim, were analysed according to the following methods. 1) Analytic rating of the instruction process Several trained observers rated 17 items according to 5 grades. The items consisted of three groups. (1) instructional materials dealt with (2) techniques of teaching (3) teacher-pupil relationships 2) Content analysis of the instruction process according to certain categories The classificatory system was based on the categories for the analysis of counseling processes and was revised especially for the present analysis of “moral” lessons. They involved (α) categories concerning the content of the problem,(β) categories concerning the function of teacher's utterances and (γ) categories concerning the pupil's utterances. Utterances were regarded as one unit when they expressed one completed idea and were classified from their psychological functions as well as from their contents, such as whether they described pupil's own conduct or the conduct of others, or whether or not they were expressed emotionally. 3) The pupil's cognition of the class (by means of a questionaire) Immediately after the class was over, the pupil were required to evaluate the class. (ex. Was it interesting? Did you learn something new? etc. The questionaire consisted of 15 items which were evaluated into 7 grades) Five months after the class, the pupils were again asked to write what he had learned in the class. The results gained through these methods were consistent with each other and seemed to reflect the characteristics of each “moral” lesson. But as for the details, not a few points are still left, to be improved. This study, we believe, played a pioneering role in studying about the effects of “moral” classes.