The scope of this paper is to present the development of a web-based evaluation system that adjusts to the level of knowledge of each student. The underlying concept is to split the questions into pools (bins) of distinct difficulty at the same time assigning a different level of accuracy to each possible answer that the student may select. Given the weighted success score computed at each round, the system raises or lowers the level of the questions asked in the following round, thus adjusting the difficulty of the test according to the students' performance. Once the score of the student improves or declines, the system again adjusts the difficulty of the questions accordingly until convergence is achieved and the students' performance becomes stable. It is believed that the proposed evaluation system predicts the level of knowledge of the students in more correct ways compared with the traditional (non-adaptive) questionnaires, while giving the students the motivation to rerun the test until they become familiar with the course content. To illustrate the evaluation methodology proposed, an online system was developed and implemented for the evaluation of sixty-eight undergraduates attending a Greek Lyric Poetry course at the Department of Classics in Aristotle University of Thessaloniki, Greece. The results of the evaluation process, with the observations made, both about the operation of the system in real conditions and the performance of the students, are also discussed in the paper. (Contains 2 tables and 9 figures.)