Multi-source evaluation of school principals is likely to become increasingly common in education contexts as the evidence accumulates about the relationship between principal effectiveness and student achievement. The purpose of this study was to examine (1) the magnitude and direction of discrepancy between how principals and their teachers perceive the principal's effectiveness and (2) what predicts principals who are at risk because their self-ratings considerably exceed the ratings others give them. We also investigated the appropriateness of various probability cut levels in analyses to predict overrating principals. The data sources were ratings by New Zealand principals (n?=?135) and their teachers (n?=?2757) of principal effectiveness--one scale (16 items) of an educational leadership practices survey. On average, both groups rated principals highly, and teachers tended to rate their principal higher than the principals rated themselves. There was more variance in teachers' ratings than principals' ratings. The variables of principal age (younger), time in principal role at the school (shorter), and socio-economic status of the school (lower) were all associated with greater magnitudes of discrepancy. Such discrepancies have implications for principals' evaluations, principal development efforts, and for school improvement.