Background and purpose Despite concerns regarding a higher risk of revision, unicompartmental knee arthroplasty (UKA) continues to be used as an alternative to total knee arthroplasty (TKA). There are, however, limited data on the subsequent outcome when a UKA is revised. We examined the survivorship for primary UKA procedures that have been revised. Methods We used data from the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR) to analyze the survivorship of 1,948 revisions of primary UKA reported to the Registry between September 1999 and December 2008. This was compared to the results of revisions of primary TKA reported during the same period where both the femoral and tibial components were revised. The Kaplan-Meier method for modeling survivorship was used. Results When a primary UKA was revised to another UKA (both major and minor revisions), it had a cumulative per cent revision (CPR) of 28 and 30 at 3 years, respectively. The CPR at 3 years when a UKA was converted to a TKA was 10. This is similar to the 3-year CPR (12) found earlier for primary TKA where both the femoral and tibial components were revised. Interpretation When a UKA requires revision, the best outcome is achieved when it is converted to a TKA. This procedure does, however, have a major risk of re-revision, which is similar to the risk of re-revision of a primary TKA that has had both the femoral and tibial components revised.