In 1989 the Medical Council of Canada embarked upon a program to incorporate into its licensing examination a clinical skills component using an OSCE. In February 1991, 240 volunteer first-and second-year residents were tested in a pilot project. The results indicated that a full-scale national administration of an OSCE for licensure was feasible with a reasonable validity and reliability. Since then three large scale administrations have now been run testing 1352, 1672, 1737 candidates. The examination consists of 20 ten minute stations, all standardized patient based using physician examiners as markers. Standard setting is done by a criterion referenced approach. Failure rates have varied from 9% to 17% for all examinees and from 4% to 7% for first time Canadian takers. Reliability estimates determined by Cronbach's alpha range from 0.72 to 0.81. Dependability indices at the cut score were 0.99 for all administrations. Generalizability analyses revealed little or no variance attributable to site. Present efforts are now being focused on sequential testing and validity studies. The Canadian experience has demonstrated the feasibility of testing clinical competence using an OSCE at a licensure level.