Until 2014 the selection for the Dutch postgraduate GP training was conducted locally, on the eight GP departments. The procedure consisted of a letter of application and a semi-structured interview. We investigated to what extend department of choice, candidates’ characteristics and qualities explained admission into GP training in a nationwide observational study of all candidates who applied in 2009/2010 (n=597). The study population addressed 542 candidates. Sixty-three candidates were rejected on letter of application. So 479 candidates were admitted to the interview, of which 340 were admitted to the GP training. Male candidates and candidates who followed medical school outside north western Europe had more risk of being rejected on application letter. Department of choice had a strong association with admission in both stages while candidates’ qualities explained admission as well. In an observational study of all candidates who entered the Utrecht selection procedure between April 2008 and 2010 (n=394), we investigated the inter-rater reliability of the interview. Twenty-six candidates (4,4%) were rejected on application letter. Ultimately, 206 of the 365 candidates (56,4%) were admitted to the GP training. The inter-rater reliability (Intraclass correlation coefficient (ICC): 0.78 - 0.84) was satisfactory. Reduction from three to two assessors slightly reduced reliability. Candidates’ qualities independently explained admission to the GP training in contrast to individual characteristics. Next, the content of a new competency based selection procedure was determined with a modified Delphi procedure. Consensus on the following CanMEDS ‘ roles was reached: ‘medical expert’, ‘communicator’, ‘collaborator, ‘manager’, and ‘professional’. Instruments were selected by searching the literature for relevant, feasible, reliable and valid methods. Four instruments were included: the National GP Knowledge Test (LHK); a Situational Judgement Test (SJT); a patterned behaviour descriptive interview (PBDI), and a ...