This study was carried out in order to identify to what extent the Turkish students' English classroom anxiety affects their academic achievement in English language. In this quantitative descriptive study, a correlational survey model was employed, and the convenience sampling was done. In order to collect data, the Foreign Language Classroom Anxiety Scale developed by Horwitz, Horwitz and Cope (1986) was used. The latest validity-reliability of this scale was conducted by Gürsu (2011), in which the three-factor structure of the scale was confirmed. The test-retest correlation of the scale was found as .85. This study was conducted with 271 university engineering students receiving compulsory prep-class education at Firat University during the academic year of 2012-2013. The data of the study were analyzed with the logistic regression analysis. According to the findings, the foreign language anxiety at the beginning of the prep-class education was not effective as the predictor of the academic achievement. On the other hand, it was observed that the students' anxiety, at the end of the education, accounted for the academic failure with its speaking anxiety in language class dimension, and predicted academic achievement with its interest towards language class dimension. This study revealed that the students' anxiety evolving and ever-increasing during their English prep-education powerfully predicted their academic performance. Whatever its source may be, anxiety in learning environments is mostly an unwanted situation. Therefore, it has always been underlined in the literature that in order to facilitate learners' language learning process, the effect of anxiety should be minimized.