This study tested two theoretical models of learner empowerment as a potential mediator of teacher power use and students' ratings of instruction. Participants included 1,416 undergraduate students from four different institutions in the United States. Results of structural equation modeling provided more support for the partial mediation model than for the full mediation model, though in the final structural model, learner empowerment failed to mediate the influence of teacher power use on students' ratings of instruction. Instead, referent, reward, and legitimate power accounted for 66% of the variance in learner empowerment, while referent, expert, and coercive power accounted for 80% of the variance in teacher evaluations. Referent power emerged as the only form of power that predicted both empowerment and evaluations. (Contains 5 figures and 1 table.)