This study reports on a review of 286 peer-reviewed articles from 2002-2018 to identify the mainstream assessment tools and strategies adopted for language proficiency assessment in computer-assisted language learning studies. Given the diversity of tools and strategies being adopted for assessment purposes, it was highly critical to trace the research trajectory in order to predict the future of assessment trends based on reported inconsistencies and suggestions. The study revealed that assessment tools and strategies addressing the dynamicity and nonlinearity of individual-test-taker differences via adaptive, interactive, and dynamic approaches were on the rise. The study highlighted reports that have been overlooked for two-decades and show the inconsistency and deficiency of adopting pencil-and-paper duplicates for computer-assisted assessment purposes. Based on the findings it can be predicted that computer-assisted assessment is moving towards integrated skills assessment as well as nonlinear dynamic individual-learner-centred assessment. Outlining the mainstream assessment tools and strategies and predicting the future direction of language proficiency assessment studies, have significant pedagogical, theoretical and assessment implications for CALL studies in particular and for the broader field of language acquisition in general.