The present study was primarily aimed at examining the applicability of Fishbein’s behavior intention theory to classroom moral instruction in the middle school. For this objective, instructional strategies and methods for moral teaching were derived from Fishbein’s theory. Those scrategic and methodological implications for moral teaching were implemented in the development of teaching plan and teacher’s instructional process. The learning units selected for the experiment were consisted of “Keeping traffic regulations”, “helping disadvantaged old men and children in the bus”, and obstructing littering and spitting”. The experimental teaching was conducted in regular 2nd grade moral classes of Banpo Middle School and Attached Middle School of Chungang University. In Banpo Middle School three different classes were sampled for three different teaching inputs; one for Fishbein’' attitudinal (Aact) treatment group, another for Fishbein’s attitudinal and normative (NB(Mc)) treatment group and the other for the control group in which conventional method of moral teaching was used. Those three methods were provided by the same teacher in charge of moral subject in the school for three classroom hours. In the Attached Middle School two groups of students, instead of three, were sampled and given two different treatments for two classroom hours: one with Fishbein’s attitudinal treatment and the other with conventional method of moral teaching. The size of each group in both schools numbered about between 60 and 70. The general scheme of experimental design used in both schools were the pretest-posttest control group method. Dependent measures adopted for the evaluation of relative effectiveness of Fishbein’s model method and the control method i.e. conventional one were attitudes, behavior intention and behavior in relation to the three respective subject areas. Data revealed that the two experimental methods i.e. Fishbein’s attitudinal treatment and his attitudinal and normative treatment were superior in their educational effectiveness compared with conventional method as measured in terms of three different measures. The relative effectiveness of Fishbein’s two methods over conventional one was appeared most saliently in attitude measures across three subject areas. This implies that the superiority of Fishbein’s methods is more likely to be reflected in attitude formation and change in classroom moral education while behavior and behavior intention components are less susceptible to change. The comparison between Fishbein’s two methods, that is, attitude-only treatment and both-attitude-normative treatment showed that the treatment on both attitudinal and normative components is more effective than the treatment on single attitude component by revealing more change in positive direction on across three dependnt meeasures. The amount of change in three measures following experimental treatment and its relative effectiveness over conventional method tended to be more prominent when provided instructional treatment for three classroom hours than when alloted two classroom hours. In the latter occasion, the effect of treatment by whatever methods, if any, tended to be markedly reduced along with elapse of time. Along with main analysis of data discribed above, subsidiary analysis were undertaken, result supported Fishbein’s assumption that behaviors tended to show higher correlations with behavior intentions than with attitudes. Most of 'the variables related to personal characteristics and familial background did not show any significant relationship to the measures of morality considered in this study except sex variable. It was concluded from the data that Fishbein’s behavior intention model might be benefitably applicable to classroom moral educational setting. However its effectiveness is more likely to be assured when classroom hour per learning unit is extended by at least one and a half time longer than the conventional classroom hour allocation. Finally more careful reexamination of Fishbein’s model in relation to educational setting is needed before the model is implemented in the field.