Background. Giant cell tumours of bone (GCT) are benign bone tumours with a locally aggressive character. Local recurrence is considered the main complication of surgical treatment and is described in up to 50% of patients. Intralesional curettage with the use of adjuvants like phenol or polymethylmetacrylate (PMMA) is recommended as initial treatment, significantly decreasing the risk of recurrence. However, risk factors for local recurrence in skeletal GCT have not yet been firmly established and a golden standard for local therapy remains controversial. Objective. The identification of risk factors predisposing for an increased risk of local recurrence. In addition, different surgical techniques are compared to identify the optimal surgical approach for the identified risk factors. Methods. In a retrospective study all 215 patients with bone GCT treated between 1964 and 2009 in one centre were included, of which 193 were suitable for analysis. All patients had minimal follow-up of 12 months (mean 115; range 12–445). Using a Kaplan Meier survival analysis recurrence free survival rates were calculated. Cox-regression was used to determine the influence of different types of therapy, the use of adjuvants, and various patient and tumour characteristics. Results. The mean local recurrence rate for all patients was 35.2% (n=68, 95%CI: 28.3–42.1). Recurrence rate after wide resection was 0.17 (n=6, 95%CI: 0.04–0.29), after curettage with adjuvants 0.32 (n=42, 95%CI 0.24–0.41) and after curettage alone 0.74 (n=20, 95%CI: 0.57–0.91, p < 0.001). Soft tissue extension (Hazard Ratio: 3.8, p < 0.001), localisation in radius and ulna (HR: 2.6, p=0.013), and surgical experience (HR: 2.2, p=0.022) were identified as significant general risk factors for local recurrence. For intralesional resection, Campanacci grade III (HR: 3.9, p=0.019) and location in axial skeleton (HR: 3.3, p=0.016) additionally significantly increased this risk. Comparing treatments our data showed that curettage followed by adjuvants was superior to curettage alone (p < 0.004), and the application of both phenol and PMMA did not present a significantly better outcome than curettage and PMMA alone (HR: 1.07, p=0.881). Conclusion. Of all possible risk factors only soft tissue extension, localisation in radius and ulna and non-radical resections significantly influenced the risk of local recurrence for all treatments. In addition, we found that high-grade tumours and localisation in the axial skeleton were additional risk factors for local recurrence after intralesional surgery. Although wide resection increases patient morbidity, it can be the therapy of choice in high risk patients. Intralesional therapy can be advised for low recurrence risk patients using curettage and PMMA only, whereas our study could not confirm the predicted effect of phenol as an additional adjuvant