Purpose Predicted survival of a patient is the most important parameter that helps to guide the treatment of a patient with metastatic spinal cancer. We aimed to investigate the reliability of modified Tokuhashi score in the decisionmaking process in patients with metastatic spinal cancer. Methods We performed a review of our prospectively collected Metastatic Cancer Database over a period of 4 years (2007)(2008)(2009)(2010). Ninety consecutive patients who were treated for metastatic spinal cancer were enrolled. Data review included demographic details, source of primary cancer, duration of symptoms, location of metastases, calculated Karnofsky's performance status, and calculated survival based on modified Tokuhashi score. We divided the patients into 3 groups. Group A included patients with expected survival less than 6 months. Group B included patients with expected survival between 6 and 12 months. Group C included patients whose expected survival was more than 12 months. We compared the calculated expected survival to the actual survival in all three groups with all patients following up to a minimum of 1 year or until death. Statistical analysis was done by Chi-square test and the Fisher Exact test.Results The survival prediction in group C was significantly accurate in 80.9 % patients (P = 0.027). However, in groups A and B, only 36.1 and 9.1 % patients survived, respectively, as per predicted. (P [ 0.05). Conclusions We can conclude from this study that, when used alone, modified Tokuhashi score may not be a reliable tool to predict survival in all patient groups.