Appointments to the specialist registrar (SpR) grade depend almost entirely on performance at interview, yet standard panel interviews do not directly assess the competences required of a medical trainee. In this study, station interviews were used to select neurology SpRs. Eighteen candidates were assessed in three interviews, each involving three stations: a curriculum vitae (CV)-based interview, an interview with a simulated patient, and a discussion of scenarios based upon teaching, audit and research. Two or three assessors at each station ranked candidates independently before discussing the pooled rankings and reading written references. The CV-based interview rankings (resembling a traditional panel interview) correlated less well with the overall rankings (r=0.54) than did research (r=0.83), information giving (r=0.75), audit (r=0.70) or teaching presentation (r=0.59). Station interviews appear fairer (providing more time, more independent examiners, fresh starts at each station), although they require more planning and expense. Competency-based assessments should be more widely used in selecting medical trainees.