This study explores the relationship between academic self-concept, classroom test performance, and causal attribution for achievement among Turkish students. 267 Year 6 students from six different cities in Turkey participated in the study. Academic self-concept was measured by means of the Myself-As-a-Learner Scale (MALS) while attributions were elicited through a specifically designed attribution questionnaire. Achievement was measured by course achievement tests. Analysis of the data revealed that teacher was the most frequent attribution for test scores followed by ability, interest, and long term effort. Academic self-concept (high vs. low) and test performance (good vs. poor) exerted a main effect but yielded together no interaction effect on attributions. Multiple regression analysis showed that a set of ability attribution, academic self-concept, interest attribution, and teacher attribution were the best predictors of test performance. As the most frequent attributions and predictors of test performance were found to be mainly uncontrollable and stable, this study concludes that knowledge of student attributions and academic self-concept can provide useful information to teachers of English both at the level of prediction and intervention beyond the level of language instruction. (C) 2014 Elsevier Ltd. All rights reserved.