Using a regional arthroplasty register, we assessed the outcome, at five years, of 1198 primary Charnley total hip replacements (THRs) undertaken across a single health region in England in 1990. An independent clinical and radiological assessment was completed for 497 operations, carried out in 18 different hospitals, under the care of 56 consultants and by differing grades of surgeon. The overall number of failures in this group was 44 (8.9%). We found that the risk of failure in patients operated on by a consultant whose firm carried out 60 or more THRs in 1990 was 25% of that of patients under the care of a consultant whose firm undertook less than 30, adjusting for a number of patient, surgeon and hospital characteristics (16% v 4%; p < 0.001 for linear trend). Our study shows that the early outcome of hip replacement surgery varies with the number of replacements undertaken by the consultant firm. A national arthroplasty register would be a convenient source for such data.