Objectives: The thesis seeks to provide an analysis of surgical work and decision making, to identify the basis of the widely observed variation in surgical practice and to indicate what surgeons see as the source(s) of that variation. Against this background, it examines the strengths and limitations of the approach promoted by the evidence-based medicine movement to surgical work. Methods: A qualitative study of surgical practice by urological and gynaecological surgeons in England and the USA involved in the treatment of female urinary stress incontinence. Depth interviews with 29 English surgeons and five American surgeons. Interviews were recorded and transcribed. Observation of 23 operations and additional ethnographic data collection at the hospitals and clinics where these surgeons worked. The observational data consist of near verbatim notes. All these data were analysed using the constant comparative approach described by Glaser and Strauss (1967). A variant of the split-half technique was used to test emerging themes. Results: Surgical practice is contingent: it is dependent on a range of variables, and, it is serendipitous. Three categories of contingency are identified (case, surgeon and external contingency). It is argued that surgical practice entails the complex interplay of these conditional factors and chance happenings. In order to learn to deal with contingency, surgeons learn or acquire practice skills through first hand experience. The thesis explores the role of the surgical apprenticeship and models of learning used by surgeons. Conclusion: The nature of surgical practice presents some fundamental challenges to EBM. The contingent and experiential features of surgical work raise serious doubts about the applicability of EBM to surgery.