Within the field of e-learning, numerous studies focus on the benefits of online delivery systems and tools for learners. Nevertheless, there is still limited understanding of why learners perform better or worse and experiments are only rarely the method of inquiry. This study reports on findings of a repetitive factorial experiment in an ecological setting with 151 secondary school pupils in order to scrutinize crucial determinants for perceived and observed benefits of two delivery modes: face-to-face versus online instruction. For these purposes, the study incorporates the DeLone and McLean's information systems success model, extended to cover perceptions of enjoyment. The findings show that pupils' performance in the e-learning condition is significantly poorer than in the face-to-face condition. The results further point to the dominant position of perceived enjoyment as a determinant of satisfaction and e-learning preference. By examining system and individual antecedents of learning performance in an experimental design, we contribute to the body of knowledge regarding online learning effectiveness. The study's limitations and opportunities for further study are discussed. [For the complete proceedings, see ED600498.]