There is a great requirement that online learning should become per-sonalized and adaptive. The paper proposes an adaptive MCQ test system which dynamically adapts to feedback of learner’s affective state during learn-ing process. Cognitive levels are associated with questions in the test. The af-fective states can be emotions like confidence or confusion.. The system identi-fies the affective state of the learner, adapts the further questions in the test se-ries in accordance with this affective state. The confused learner will get easier questions and confident learner will get more difficult questions. This process will continue for the learning activity. Thus test system becomes adaptive.