BACKGROUND Unicompartmental knee arthroplasty has been associated with consistently worse implant survival rates than total knee arthroplasty in worldwide arthroplasty registers. The rate of revision and the proportion of revisions performed for unexplained knee pain after either a unicompartmental or total knee arthroplasty were studied to assess if there is evidence to support the hypothesis that the numbers of revisions performed for unexplained knee pain differ between these two implant types. METHODS Using data from the National Joint Registry (NJR) of England and Wales, we identified 402,714 primary knee arthroplasties (366,965 total knee arthroplasties and 35,749 unicompartmental knee arthroplasties) that were consecutively entered from April 2003 to December 2010. The status of all implants was assessed as of December 2010, at which time 6075 implants (4503 total knee implants and 1572 unicompartmental knee implants) had been revised at a maximum of eight years. Survival analysis and Cox regression analysis with adjustment of differences in age, sex, American Society of Anesthesiologists (ASA) grade, and indication for arthroplasty were performed with use of the end points of revision for any reason, revision for unexplained pain, and revision for other reasons. RESULTS Revision for unexplained pain was more common after unicompartmental knee arthroplasty than after total knee arthroplasty (representing 23% of revisions as compared with 9% of revisions; p < 0.001). The five-year rate of revision for unexplained pain was 1.6% for the unicompartmental knee arthroplasty group and 0.2% for the total knee arthroplasty group. With use of Cox regression, the hazard ratio (HR) for unicompartmental knee arthroplasty relative to total knee arthroplasty with the end points of revision for any reason, revision for unexplained pain, and revision for all other reasons were 2.82 (95% confidence interval [CI], 2.66 to 2.99; p < 0.001), 6.76 (95% CI, 5.84 to 7.83; p < 0.001), and 2.39 (95% CI, 2.24 to 2.56; p < 0.001), respectively. The mean time from primary implantation to revision was similar for both implant types. CONCLUSIONS While more unicompartmental knee implants than total knee implants were revised for unexplained pain, when these revisions for unexplained pain were discounted, unicompartmental knee arthroplasty still had a significantly greater risk of revision from other reasons than did total knee arthroplasty. The revision rate in isolation may not be a reliable way to compare different implant designs and should instead be considered in the context of the reason for failure.