Background <p>Evaluation of clinical performance is essential in all medical school programmes. Students undergo multiple clinical placements in different disciplines and settings, and typically must pass an end-of-placement supervisor sign-off evaluation to progress. However, the validity of this sign-off model remains unclear. This study aims to assess the extent to which this assessment method predicts performance in summative medical school examinations.</p> Methods <p>We compared summative knowledge and clinical skills examination scores with end-of-placement supervisor sign-off ratings of ‘knowledge’, ‘clinical skills’ and ‘practical skills’ for medical undergraduate students, across three clinical placements at Imperial College London, UK (n = 355). Statistical analysis for predictive validity was performed through Ordinary Least Squares regression.</p> Results <p>End-of-placement supervisor ratings in hospital did not significantly predict student performance in summative knowledge tests or clinical skills assessment. ‘Knowledge’ and ‘practical skills’ ratings lacked predictive validity across all supervisors. Statistically significant predictive validity was evident for GP supervisor ratings of ‘clinical skills’ and examination scores, but the effect size was educationally insignificant (p = 0.01, r² = 0.02).</p> Conclusions <p>End-of-placement supervisor ratings did not demonstrate educationally significant predictive validity towards end-of-year examinations. Multi-source feedback, embedded in-placement assessment, and additional formalised supervision time in supervisors’ work schedules could be beneficial to improve the educational value for the student and the clinical placement sign-off process. Different sign-off requirements could be considered for GP and hospital settings, with tailoring of constructs to suit the clinical environment.</p>