This study compared 96 high efficacy and 93 low efficacy elementary teachers regarding a variety of cognitive, affective, and behavioral factors associated with classroom management and discipline of at-risk students. Quantitative and qualitative data were collected in relation to: teachers causal attributions for student misbehavior, expectancy for student behavior improvement, emotional reactions to student misbehavior, student control ideology, and recalled frequency of using restrictive and positive intervention strategies. A series of analysis of variance procedures indicated that high efficacy teachers were less likely to judge their difficult students as having chronic behavior problems; more likely to expect student behavior improvement; less likely to feel angry, embarrassed, or guilty about student misbehavior; more likely to like problem students; and more likely to feel confident about being able to manage misbehavior. Path analysis supported the development of a model in which low teacher efficacy, mediated by anger and stress, predicts usage of severe punishments. The study showed support for the construct of teacher efficacy as an important marker of general teacher effectiveness in the domain of classroom management. The questionnaire is appended. (Contains 99 references.) (SM)