Edgar is a web application being developed at the Faculty of Electrical engineering and Computing (University of Zagreb), for use in educational institutions with the need for on-line test assessment and examination. Within this master’s thesis, the set of features in Edgar has been expanded with the support of on-line learning experiences in the form of tutorials and adaptive exercises. Tutorials present course material in sequentially arranged steps, assigning one question per step associated to the presented material. Based on a particular student’s behaviour in the tutorial, appropriate hints are shown to provide help with the understanding of material, knowledge application within exercises. Adaptive exercises apply adaptive learning principles by assigning questions of appropriate difficulty level based on the student’s success rate with previous questions. Students are also left with the possibility to pick the desired difficulty level for further questions. The difficulty level of questions is determined through analysis of the results of previously graded tests that contained those questions. Question difficulty is estimated using psychometric methods of the item response theory. Latent trait models from IRT are used to determine the parameters of item characteristic curves. The ICC shows the probability of giving correct answers for subjects of different ability levels. Teachers can automatically assign final question difficulty levels using several available calculated difficulty models. Final question difficulty can also be manually assigned by the teacher. Individual and collective statistical question attempt and score indicators have been implemented for tutorials, adaptive exercises and tests in both table and chart forms. Students have been given the ability to leave feedback in textual and grade form. Teachers have been given the ability to filter and browse feedback based on questions, tutorials, exercises, students and class groups.