The Palliative Prognostic Index (PPI) has recently been developed for survival prediction in terminally ill cancer patients. To clarify whether physicians' clinical prediction of survival (CPS) can be improved using this objective assessment aid, two sequential prospective studies were performed on two independent series of hospice inpatients (n = 150 and 108). In the first study, the CPS was prospectively recorded by primary physicians on the basis of their clinical experiences. In the second study, physicians estimated patient prognoses with a reference to the PPI score. The cases where the differences between actual survival (AS) and CPS were 28 days or longer significantly decreased in the second study compared to the first study (42% vs 23%, P < 0.01). Also, the cases where AS was either twice longer or half shorter than CPS significantly declined (49% vs 37%, P = 0.050). As well, serious errors, defined as the cases where AS was either (a) 28 days and twice longer than CPS or (b) 28 days and half shorter than CPS, significantly decreased from 27% in the first study to 16% in the second study (P = 0.028). In conclusion, the PPI can contribute to improvement in physicians' ability to predict survival of terminally ill cancer patients.