This article puts forth a multidimensional framework for empirically testing the effects of teacher mentoring, focusing on interactions between formal mentors and novice teachers and the setting in which these interactions occur. Analyzing survey and administrative data from Chicago Public Schools with a multilevel propensity-score weighting approach, our analyses identifies the combination of stronger leadership and high-quality mentoring that features at least biweekly mentor-mentee interactions, comprehensive content, and opportunities for engagement with teaching practice as the most effective for building teacher organizational commitment. The results also reveal that stronger principal leadership may protect teachers from lack of access to any mentor or to high-quality mentoring while weaker leadership may diminish the potential benefits of high-quality mentoring.