Acetabular reconstruction using an ice-cream cone prosthesis has been a reliable reconstruction option following pelvic tumour resection. However, it remains unknown which factor determines the success of this procedure. We aimed to determine risk factors for complications and functional loss in acetabular reconstruction using an ice-cream cone prosthesis. Fifty-four patients with malignant bone tumours who underwent acetabular reconstruction using an ice-cream cone prosthesis between 2004 and 2016 were studied. The bone-stem ratio was calculated as the ratio of the inserted length into the bone per the entire stem length. A total of 26 (48%) patients had at least one complication and 11 patients (20%) required surgical interventions. The complication rates were 71% and 40% with a bone-stem ratio ≤ 50% and > 50%, respectively (p = 0.026), and the bone-stem ratio significantly stratified the risk of complications (≤ 50%: OR, 4.67 versus > 50%; p = 0.048). The mean MSTS score at the final follow-up was 60% (range 23-97%): the scores were significantly lower in patients with complications/leg-length discrepancy (52%) than in those without (79%; p = 0.002). The mean score with a bone-stem ratio ≤ 50% was significantly lower than the score with a ratio > 50%, especially in patients who underwent non-navigated reconstructions (33% versus 64%; p = 0.001). The inserted length of the coned stem into residual bone was predictive of complications and functional outcome. Surgical indication for this procedure should be considered with the size of the remaining ilium to stabilise the prosthesis with a coned stem longer than half length.