The Swedish National Total Hip Arthroplasty Register records primary hip replacements, revisions and surgical technique/environmental factors. The end-point for failure is revision. A prosthesis still in place, however, does not mean success. Clinical and radiographic outcomes should describe in more detail the efficacy of hip replacement surgery instead of the relatively blunt outcome measure that the register can provide. We performed a clinical outcome analysis on patients with primary total hip replacement thus testing the adequacy of the end-point for failure in the Swedish register. 1,113 randomly selected patients who had had total hip replacement surgery between 1986 and 1995 answered a disease-specific self-administered questionnaire (WOMAC). A cohort of 344 patients was studied, using the Harris Hip Score and a conventional radiographic examination as outcome measures. We found clinical failure rates of 13% and 20% for all implants after 10 years, using 60 points or revision as the definition of failure in the Harris Hip Score and WOMAC, respectively. The result, according to the register during the same period, was a 7% revision rate. The clinical failure rate depended on the type of evaluation tool, definition of failure and demographics, which made it difficult to decide whether there was a need for revision. With the exception of pain measured by the Harris Hip Score, the results showed no significant correlation between clinical failure and radiographic failure. Hence, with the knowledge that there is a difference between the revision rate according to the register and clinical outcome, the strict definition of failure in the register is useful as an end-point for primary hip replacement surgery.