Background: Patient selection for either total knee arthroplasty or unicompartmental knee arthroplasty remains controversial. The latter has several reported advantages over total knee arthroplasty, but it also appears to have significant drawbacks in terms of revision rates. Aims: This study aimed to determine the influence of the preoperative degree of osteoarthritis on the risk of reoperation following unicompartmental knee arthroplasty. Methods: Surgery was carried out on 294 knees in 241 patients between 2001 and 2012 at a single institute, using cemented Oxford phase III unicompartmental knee arthroplasty. The mean age at the time of operation was 67 years, and the mean follow-up time was 8.7 years. Results and Conclusion: The knees with a preoperative Kellgren–Lawrence grade of 0–2 osteoarthritis had a higher risk of reoperation than those with a Kellgren–Lawrence grade of 3–4 (odds ratio = 1.89; 95% confidence interval, 1.03–3.45; p = 0.04). In addition, the knees with a medial joint space width of more than 1 mm or a high medial/lateral joint space width ratio had an increased risk of reoperation. In conclusion, we suggest that unicompartmental knee arthroplasty should only be performed in cases showing severe osteoarthritis in preoperative radiographs, with medial bone-on-bone contact, and a medial/lateral ratio of <20%.