Background: In problem-based learning (PBL) curricula implemented around the world, it is assumed that students adopt a deep learning approach to studying and aim to gain a profound understanding of the subjects being studied. However, it is not clear which PBL components initiate or deter deep learning and to what extent this happens and why. Aim: This study explored to which extent students used a deep or surface learning approach in PBL and whether this differs across years. We also investigated which PBL components students perceived to be hindrances to deep or surface learning. Methods: The study took place at Sulaiman Al Rajhi Medical College, Qassim, Kingdom of Saudi Arabia. A mixed-methods approach was applied. A validated questionnaire and semi-structured focus group interviews were conducted sequentially. Results: First-, second-, and third-year students reported, in scale 1–5, for deep learning scores, respectively, with mean (M) = 3.55, M = 3.41, and M = 3.55. First-, second-, and third-year students reported, in scale 1–5, for surface learning scores, respectively, with M = 2.88, M = 2.78, and M = 2.89. The differences for both deep and surface learning across the years were statistically nonsignificant. According to students, they study deeply on main learning objectives and superficially on minor objectives as indicated by tutors, they are stimulated toward deep learning through interesting topics during self-study, and examinations drive them toward deep or surface learning depending on the question format and necessity to pass. Conclusions: The results of this study confirm that students' perceptions of PBL components affect their approaches to deep and surface learning. These effects are not entirely negative or positive. Students seem to frequently employ a deep learning approach in PBL throughout the 3 years. These conclusions will allow program administrators/educationalists to constructively design curricula around the perceptions of learners of PBL tutors, topics, and examinations.