Conflicting evidence as to the presence or absence of generalization in classroom behavior modification programs prompted this study of the conditions of generalization. During the experiment, behaviors operationally defined as competitive or cooperative were reinforced in certain game situations. Then the generalization of this training over variables of task, type of response, and time periods was measured. It was predicted that generalization of the reinforced response would occur most strongly in the testing situation most like the original one. Experimental results confirmed these expectations. Results showed in addition that there was no significant difference in competitiveness between boys and girls. (RB)