The constant public demand for high-quality medical services drives an associated demand for professional doctors, and requires them to take high-stakes exams. MRCP(UK) is a major examination for physicians in the UK, which aims to assess their knowledge, skills and appropriate attitudes – all key aspects of being a medical professional. Although the existing literature provides extensive evidence supporting the quality of MRCP(UK), this research aimed to add to the existing body of knowledge by investigating the predictive validity of MRCP(UK), i.e. examining whether it truly selects candidates who possess the above-mentioned qualities. This research therefore investigated the relationships between MRCP(UK) scores and results of seventeen knowledge exams and two clinical skills assessments (including specialty exams and MRCGP), training performance assessment outcomes (ARCP), and cases of licence limitations and erasures. Operating with the hypothesis that MRCP(UK) would predict all of the above-mentioned criteria, a retrospective longitudinal approach was assumed. The main sample contained records of 50,311 MRCP(UK) candidates attempting MRCP(UK) between May 2003 and January 2011; however, the analyses were performed on smaller samples, from 8 to 33,359 cases, depending on the size of the criterial dataset. The results of univariate and multivariate analyses supported the hypothesis. MRCP(UK) scores were indeed predictive of results of all knowledge exams and clinical assessments (meta-analysed average effects: r=0.69 for Part I, r=0.70 for Part II, 0.48 for PACES), and of performance in specialty training and issues with the licence to practice (on average: r=0.24 for Part I, and r=0.22 for Part II and PACES). The magnitudes of these validity coefficients were consistent with the theoretical notions of psychometrics and concurred with the findings of published studies. In view of the evidence it was concluded that MRCP(UK) is a valid exam. The limitations of this study, directions for future research, and general implications were discussed.