S ummary . Ninth grade students made attributions for failing tests in liked versus disliked school subjects and indicated the attributions their teachers would make for their failure outcomes. Students employed, and expected that their teachers would employ, different patterns of causal attribution to explain their failure in liked versus disliked subjects, with the former yielding more test‐specific attributions. Differences were found between students' attributions and the attributions they expected their teachers to make for the same failure outcomes. The data provide evidence for the important role of ego‐involvement, as indexed by liking of school subjects, in the causal attribution process.