Burnout is conceptualized and measured as a three-dimensional construct consisting of emotional exhaustion, cynicism, and personal efficacy. Research suggests that the personal efficacy scale be replaced by a personal inefficacy scale. The purpose of this study was to explore (a) differences in the general factor saturation of the Maslach Burnout Inventory–Student Survey when efficacy or inefficacy items are included, (b) the strength of the correlations of the efficacy versus inefficacy scales with emotional exhaustion and depersonalization, and (c) the reliabilities of the different subscales. Two unrestricted maximum likelihood item factor analyses with bifactor rotations were conducted on data collected from 522 South African university students. The results indicate that the inefficacy scale loads more strongly on a general burnout factor than does the efficacy scale, the inefficacy scale correlates more strongly with the exhaustion and cynicism scales, and the inefficacy scale is more reliable than the efficacy scale. Jointly this implies that the inefficacy scale, rather than the efficacy scale, should be used in the measurement of burnout.