The survival outcome of patients with systemic cancer differs significantly between individuals even within the same tumour type. We set out to illustrate this by analysing the factors determining survival in patients with metastatic disease from nasopharyngeal carcinoma (NPC) and to design a scoring system based on these prognostic factors. Patients referred between January 1994 and December 1999 were retrospectively analysed. Factors analysed included patient (age group, gender, performance status (BS) at diagnosis of metastases), disease (number of metastatic sites, specific metastatic sites, disease-free interval (DFI), metastases at presentation, presence of locoregional recurrence), and laboratory factors (leucocyte count, haemoglobin level, albumin level). Univariate and multivariable analyses were performed using the Cox proportion hazards model. A numerical score was derived from the regression coefficients of each independent prognostic variable. The prognostic index score (PIS) of each patient was calculated by totalling up the scores of each independent variable. Independently significant, negative prognostic factors were liver metastasis, lung metastasis, anaemia, poor PS, distant metastasis at initial diagnosis, and a DFI of <6 months. Three prognostic groups based on the PIS were obtained: (i) good risk (PIS=0-6); (ii) intermediate risk (7-10); (iii) poor risk (>or=11). The median survivals for these groups were 19.5, 10, and 5.8, months, respectively, (log rank test: P<0.0001). The variable prognosis of patients with disseminated NPC can be assessed by using easily available clinical information (patient, disease and laboratory factors). The PIS system will need to be validated on prospectively collected data of another cohort of patients.