This study analyzes teacher efficacy for fourth- and fifth-grade mathematics teaching and its relationship to teacher and student performance as measured through classroom observation ratings by administrators and mathematical growth of the students. Quantitative correlation methods were used, including point-biserial regression, single regression, and multiple regression. The subjects of the study were 32 teachers in a single school district in Florida. The theoretical framework for the study is rooted in Bandura's construct of self-efficacy as measured by two different instruments: one measuring general teacher self-efficacy and a second measuring mathematics teaching efficacy beliefs. Teacher observation results were based on scores from the Marzano teacher evaluation model, and student growth was measured using Florida value-added mathematics scores for each teacher. Teacher evaluations were based on a combination of observation scores and value-added scores. Significant relationships were identified that validate Bandura’s triadic reciprocal framework.