Virtual patients are interactive tools commonly used by medical schools for teaching and learning, and as training tools for the development of clinical reasoning. The feedback delivered to students is a crucial feature in virtual patients. Personalised feedback, in particular, helps students to reflect on their mistakes and to organise their knowledge in order to use it appropriately in a clinical context. However, authoring personalised feedback in virtual patient systems can become a di�cult task, due to the large number of choices available to students and the complex implications of each choice. Additionally, the current technologies used for the design and exchange of virtual patients have limitations in terms of interoperability and data reusability. Semantic web technologies are designed to model complex knowledge in a flexible manner, allowing easy data sharing from multiple sources and automatic data processing. This thesis demonstrates the benefitts of Semantic Web technologies for the design of virtual patients, in particular for the automatic generation of personalised feedback. Seven important types of personalised feedback were identified from the literature, and a preliminary survey showed that students in year 3 to 5 consider two of these types of feedback to be particularly useful: feedback indicating actions that each student should have chosen but neglected, and feedback indicating the diagnoses that each student should have tested and rule out or confi�rmed, given the initial presentation of the patient. SemVP, a Semantic Web-based virtual patient system, was created and evaluated by medical students, using a quantitative survey and qualitative interviews. This study showed that SemVP can generate useful personalised feedback, without the need for a virtual case author to write feedback manually, using a semantic model representing both the virtual patient and each student's actions, and leveraging existing data sources available online.