The purpose of this study was to provide the surgeons with effective and reliable guidelines for surgical decision-making by establishing a scoring system for giant cell tumour (GCTSS) based on evidence and expert opinion. The modified Delphi technique and analytic hierarchy process were used to establish the GCTSS. The GCTSS was defined and classified based on different surgical methods using data from 207 patients collected retrospectively between October 2003 and December 2014. Finally, prospective data of 40 patients between December 2014 and October 2015 were used to analyze concordance between score categorization and experts' consensus on surgical procedure. A novel GCTSS included pathological fracture, cortical bone destruction, tumour size, and articular surface involved. The total scores ranged from 1 to 12 points. The strategy for each patient was decided: a total score of 1-4 suggested intralesional curettage alone for excellent post-operative function; 5-9 points indicated intralesional curettage with internal fixation for less surgery-related complications; and 10-12 points indicated prosthesis replacement for long-term local control. The κ-statistic for the predictive validity of total score was 0.611. The κ coefficient of each group represented moderate or substantial agreement, which was acceptable. The intraclass correlation coefficient for inter- and intra-observer reliability of total score was 0.831 and 0.740, respectively. The novel GCTSS is a comprehensive scoring system with content validity that can aid surgeons in assessing the aggressiveness or severity of giant cell tumour and might become a prognostic tool for surgical decision-making.