This paper presents a study that examined the effect of an attribution induction on students' performance on a subsequent achievement test. The study was conducted to determine whether an instructor might inculcate desirable attributions as part of a regular classroom interaction and thus improve academic performance in at-risk students. College students (N=263) at the University of Manitoba (Canada) were informed that either effort, ability, or test difficulty was the primary determinant of their performance on a test. They then viewed a videotaped lecture presented by either a low or high expressive instructor. Following this, they wrote an achievement test based on the lecture content. Results showed that with low expressive instruction, effort attributions enhanced achievement, but only for students who perceived themselves as successful on the first test. Attributions did not affect the performance of less successful students when instruction was ineffective. Moreover, high expressive instruction improved achievement with students performing well regardless of the attributions induced. It was concluded that it may be useful to consider attributional retraining as part of regular teaching activities. Contains 19 references. (Author/GLR)