After the hype of using computers in the 2000's in pure e-learning environments, the results have been disappointing. Blended learning is suspected to combine the benefits of traditional courses with e-learning. Aims: What are student's perceptions of their learning experience in traditional courses, e-learning and blended-learning groups? Is there an influence on the test results afterwards? Materials and Methods: 75 students attending a orthodontic course in diagnostics (facial diagnostics) were divided into three groups, traditional lecture and seminar (A:n=26), e-learning (B:n=11) and blended-learning (C:n=38). The students had no prior knowledge to the facial analysis used. Special e-learning software for facial analysis was developed (Borland Delphi 7, including patient management and facial analysis). Each student was evaluated with a questionnaire with 20 items in four categories (motivation, didactics, response and effect) after completing the course. In addition each student had to pass a test with 20 images to analyse afterwards. Results: The questionnaires showed that blended learning leads to a more active role of the student (C:77% vs B:33% and A:35%) and the perceived increase in knowledge was also higher in group C. Group differences were tested with Kruskal-Wallis-test and multiple pair comparison (Dunn). Group B was rated inferior or at least equal from most students in all categories. In the category supposed effect group B is significantly inferior to all other groups. In the test there was been no significant difference between the groups. Conclusions: As a result e-learning only cannot be recommended in teaching facial analysis in orthodontics. Regardless of the content e-learning may be at best equal to a lecture. The networking of electronically generated content and personal contact leads to higher motivation but there can be no better test results expected. Keywords: e-learning, blended learning, orthodontics