This article reports on a mixed-method study that examined the washback of a local integrated theme-based high-stakes English language proficiency test that is used in a university English for Academic Purposes (EAP) program in Turkey. The assumption behind employing an integrated theme-based test, which resembles authentic language use, was that it would bring about a positive washback on learning (Leki, Cumming & Silva, 2008: Leki & Carson 1997). The data were collected from both focus-group interviews after the instruction and pre- and post- proficiency test scores of 147 EFL students in the Preparatory English Language Program (PEP). Test of Readiness for Academic English (TRACE) was administered at the beginning and at the end of a 4-month English language instruction period. Repeated measure ANOVA and inductive analysis of the transcribed interview data were used for analyzing quantitative and qualitative data respectively. The findings indicated that the test had both positive and negative washback on the learning. Most students considered that using source-based information and their notes taken during the listening task into their writing raised their awareness in terms of generating, organizing and linking ideas as well as modelling vocabulary and sentence structures. However, the test also exerted negative washback upon learning since students were inclined to prioritize test-oriented practice. The implications of the study suggest that a theme-based integrated proficiency exam may elicit positive washback on learning that could be used for validity evidence in EAP contexts and lead to more appropriate language assessment. The procedures are detailed, the findings are presented and discussed, the applications and implications for teachers and test designers are explained, and some suggestions are made for further research.