Summary This mixed-method study examined how task complexity and collaborative learning influence students’ cognitive and emotional responses during simulation-based training with virtual patients (VPs). Seventy-five dental students were randomly assigned to one of four training conditions in a 2×2 factorial design: low-complex/individual, low-complex/collaborative, high-complex/individual, and high-complex/collaborative. Emotional states were measured before and after VP sessions, and cognitive load was assessed during training. In the assessment phase, 67 students completed two new VPs of varying complexity. Additionally, twelve students were interviewed about their emotional experiences. The results revealed significant interactions between task complexity and learning mode, particularly in affective outcomes: students trained collaboratively with low-complex VPs reported higher levels of invigoration than those in the high-complex condition. The findings underscore the importance of considering both cognitive and affective factors in the instructional design of collaborative VP training. Dataset The dataset includes both quantitative and qualitative data from 75 participants. The first file, ‘emotional and cognitive measures’, contains anonymized participant data on emotional state ratings (before and after each VP), self-reported cognitive load, task complexity, and learning mode. The second file, ‘interview excerpts’, includes coded quotes from twelve students describing their emotional experiences with VPs, categorized by training condition and emotional themes (e.g., invigoration, anxiety). This dataset supports the exploration of how instructional design features influence students’ emotions and cognitive processing in simulation-based dental education.