Considerable efforts have been made to predict success in medical degrees. Much of the work has focused on failing students, so little is known about performance stability in medical students who pass and become doctors. If we can predict performance, we can better plan interventions and set standards. We tested the predictive capability of first year assessment on final year assessment marks in 314 graduating medical students. A linear regression model showed around half the variance in final year performance is explained by first year performance despite the very different nature of early assessment. Marks at graduation can be predicted with some accuracy using only first year results. With this information, we can better anticipate student performance and set defensible passing standards. Either first and final year assessment measures the same underlying attributes (cognitive ability, conscientiousness), first year assessment provides an absolutely critical foundation for graduation, or both.