To foster medical students' diagnostic knowledge a case-based worked example approach was implemented in the context of a computer-based learning environment. Thirty medical students were randomly assigned to the condition "with erroneous examples", and 31 students learned with correct examples. Diagnostic knowledge was operationalised using a multiple-choice test, key-feature problems and problem-solving tasks. Cognitive load, acceptance and subjective learning outcomes were assessed by three questionnaire scales. Students learning with erroneous examples performed better than students learning with correct examples. However, this effect was not significant. Erroneous examples significantly increased cognitive load and decreased students' subjective learning outcomes. However, they had no influence on acceptance. (Contains 1 table.)