The objective of the study was to examine the effectiveness of an intervention based on attribution retraining with regards to student misconduct and coercive teacher behavior. An intervention would lead to a sustained decrease in misbehavior and coercive discipline without using any external control systems. In this case study, a male, veteran Grade 8 teacher and his students were involved in a long-lasting conflict characterized by an increase of disruptive student conduct and the teacher's coercive behavior. Inspired by the explanatory model of the extended symmetrical escalation (Lapointe, 2003), the researchers developed an intervention using attribution retraining to resolve the problem. Data collected from observations, questionnaires, and interviews indicated that this non-coercive intervention contributed to long-term general improvement of teacher-student relations and behaviors. The results highlight the importance of perceptions within conflicting interactions and lead to recommendations for interventions that could be used to reduce group discipline problems.