Abstract: The aim of this pilot study was to investigate factors which may be predictive of final examination results obtained by first-year medical students in a graduate-entry, problem-based learning (PBL) course. Using pre-tested and novel instruments, key measures were taken in the seven months preceding final year examinations, including: self-efficacy for self-regulation of learning in PBL; self-efficacy for academic achievement; approaches to learning - deep, strategic, and surface; reflection on learning; and tutor assessment scores of students' critical thinking abilities. In addition, final written and clinical examination results were obtained. Data analysis revealed statistically significant correlations between final examination results and clinical exam scores, self-efficacy, deep learning, and tutor assessment of students' critical thinking abilities. Statistically significant negative correlation was demonstrated between final examination results and surface learning. Forward regression showed that three of these factors predicted 46% of the variance in student examination results. These were clinical exam scores, tutor-assessment of critical thinking, and self-efficacy for self-regulation. Clearly, academic performance results from a complex and inter-related set of knowledge, skills, attitudes and perceptions. This pilot study provides evidence of the importance of using diverse methods to try to identify those students less likely to achieve success so that they may be offered learning support and guidance in the months leading up to their final examination. Furthermore, it suggests that tutors' role in identifying students at risk of poorer academic performance needs to be more thoroughly researched.