Accurate diagnosis of racing injuries can be difficult. The objectives of this study were to describe the postmortem (PM) defined distribution of fatal distal limb fractures (DLF) affecting Thoroughbreds racing in Great Britain between February 1999 and August 2005 and then assess the accuracy of veterinary racecourse diagnoses and examine whether these improved following introduction of a computerised recording system. PM examinations were performed on limbs distal to radius or tibia from all cases of fatal DLF occurring on British racecourses during the study period. Results of these examinations were described and compared with the diagnoses made at the racecourses. Over the study period, fatal DLF prevalence in all race types was 0.63 per 1000 starts (344/545,335), with the lowest frequency (0.34 per 1000 starts) in flat racing on turf and the highest frequency (1.56 per 1000 starts) in national hunt flat races. The prevalence of fatal DLF in steeplechase racing had reduced from that reported previously: from 2.3 to 1 per 1000 starts. Racecourse veterinary identification of fracture presence was good (>93 per cent); however, identification of all fractured bones was poor (<55 per cent). Introduction of a computerised recording system did not significantly improve diagnostic accuracy. The prevalence of fatal DLF has not significantly altered since the 1970s. Techniques such as on course digital radiography to help improve racecourse fracture diagnoses could be introduced.