In the past two decades, corpora have been proposed as valuable computer-assisted tools for teaching and learning academic writing in English at the university level. This article reports on an empirical study that sought to examine the effectiveness of direct corpus consultation in overcoming vocabulary errors in academic writing. This study is part of a larger research project that also investigated students' attitudes towards direct corpus consultation activities in terms of overcoming vocabulary errors and promoting their learning autonomy as well as the difficulties they encountered. A quantitative quasi-experimental design was used in the present study, which involved 32 Saudi female undergraduates majoring in English for nine weeks of the fourteen-week writing course. Data collection instruments included a pre-post vocabulary test and pre-post writing tests. The results of quantitative and content analyses suggested that direct corpus consultation decreased vocabulary errors and improved students' academic writing quality. The corpus consultation tasks significantly affected students' ability to overcome the four targeted types of vocabulary errors (collocation, connotation, word form, and spelling). Furthermore, the results showed that the effectiveness of corpus consultation tasks in improving students' vocabulary knowledge and usage varied depending on the type of vocabulary error. On the vocabulary test, collocation errors had the smallest decrease and medium effect size, whereas, on the writing post-test, connotations were the least developed type. The study concluded with some pedagogical implications for using direct corpus consultation to develop students' vocabulary knowledge and usage, as well as recommendations for future research.