Numerous studies have found that blame for failure can have a „seemingly paradoxical“ effect on ability attributions. This effect consists in the inference that the blaming person estimates the blamed person’s ability as high. In a recent study this effect could no longer be replicated for the total group of subjects, but occurred only for a subgroup, whereas the remaining subjects showed an opposed attributional pattern. Empirical evidence is furnished that the subjects showing versus not showing the paradoxical effect used different causal schemas for the interpretation of teacher blame for failure. These schemas were based on the perceived estimate by the teacher of the blamed student’s effort, versus on his perceived liking vs. dislike for the student, respectively.