Choosing inappropriate or unethical actions in games is referred to as dark play. For a serious game on delirium for medical students, we aimed to investigate the potential differences between dark play and normal play on game effectiveness regarding abilities in advising care, learning motivation and engagement, and attitude toward delirious patients. Furthermore, we aimed to explore the use of different game features between the two types of play on empathy, self-efficacy, and consequences of care. We performed a two-arm randomized controlled trial including an exploratory qualitative approach with 157 medical students, who played the serious game "The Delirium Experience." Participants were randomly allocated to either the dark play or normal play group. Participants had to give three recommendations for taking care of delirious patients, and complete both the Delirium Attitude Scale, and Learning Motivation and Engagement Questionnaire to study game effectiveness. To explore game features, open questions were asked. We did not find difference between the two types of play in game effectiveness. Patient's and nurse's perspective" seem to be an important game feature for being able to empathize with a patient in both groups. To support self-efficacy, "practice how to care" and "feedback in the game" were important in both study groups. "Being able to see the importance of good interaction with the patient" was reported important for self-efficacy in the dark play group, whereas this was "seeing the consequences of care" in the normal play group. There seems to be no change to game effectiveness when providing players the opportunity to use dark play in a serious game. A realistic view of another person's perspective could be an important game feature to increase empathy.