Purpose: Indications for unicompartmental knee arthroplasty (UKA) vary between units. Some authors have suggested, and many surgeons believe, that medial UKA should only be performed in patients who localise their pain to the medial joint line. This is despite research showing a poor correlation between patient-reported location of pain and radiological or operative findings in osteoarthritis. The aim of this study is to determine the effect of patient-reported preoperative pain location and functional outcome of UKA at 1 and 5 years. Methods: Preoperative pain location data were collected for 406 knees (380 patients) undergoing Oxford medial UKA. Oxford Knee Score, American Knee Society Scores and Tegner activity scale were recorded preoperatively and at follow-up; 272/406 (67 %) had pure medial pain, 25/406 (6 %) had pure anterior knee pain, and 109/406 (27 %) had mixed or generalised pain. None had pure lateral pain. The primary outcome interval is 1 year; 132/406 patients had attained 5 years by the time of analysis, and their 5-year data are presented. Results: At 1 and 5 years, each group had improved significantly by each measure [mean ΔOKS 15.6 (SD 8.9) at year 1, 16.3 (9.3) at year 5]. There was no difference between the groups, nor between patients with and without anterior knee pain or isolated medial pain. Conclusions: No correlation is demonstrated between preoperative pain location and outcome. We conclude that localised medial pain should not be a prerequisite to UKA and that it may be performed in patients with generalised or anterior knee pain. Level of evidence: Therapeutic, Level II. © 2012 Springer-Verlag.