In our hospital all operative procedures are coded using the OPCS 4.3 classification and in addition are entered into an independent theatre databases. Using these two databases we identified patients undergoing hip fracture surgery at this hospital between 1st November 2003 and 30th November 2006. We identified 408 cases. No single database identified all 408 cases. A quarter of cases (N=98) were not procedurally coded. Only 43.2% (N=176) of cases were recorded in both the theatre database and procedurally coded at the time of this study. Overall the coding accuracy of these 176 cases was 93.8%. Clinical coding at this hospital was unreliable and inaccurate, which has major implications for national statistics, performance analysis and most importantly Payment by Results. We discuss this further and offer possible solutions to improve the coding process.