Introduction: The importance of estimating the prognosis of advanced cancer patients is well known, but clinicians do not estimate survival time accurately. Since there is a need for an objective index to estimate survival time, the utility of the Prognostic Nutritional Index (PNI), which depends only on objective factors, was evaluated. Methods: The PNI was calculated using the following formula, PNI=10×serum albumin value (g/dL)+0.005×lymphocyte count in peripheral blood, at 3 months, 2 months, 1 month, 3 weeks, 2 weeks, 1 week, and within 3 days before death in 278 cancer patients (166 men, 112 women; age range, 33-99 years; mean age, 69.8 years) who died in a hospital surgical unit. Results: Sites of primary diseases included lung, breast, esophagus, stomach, colorectum, liver, biliary tract, and pancreas. The PNI values showed a gradual decrease over time. Changes in the PNI values were lower in non-gastrointestinal cancer patients than in gastrointestinal cancer patients. The mean PNI value was significantly higher in patients who lived >3 weeks (38.8) than in those who died within 3 weeks (32.4). When the PNI cut-off point was set at 35, and it was assumed that the life expectancy was within 3 weeks in cases with PNI <35, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 74.8%, 62.2%, 68.1%, and 69.6%, respectively. Discussion: The PNI appears to be a useful and simple parameter to predict clinical outcomes of patients with terminal stage cancer. Particularly, the PNI is considered feasible for gastrointestinal cancer patients.