In a prospective open study, 61 consecutive patients with advanced cancer admitted to a Palliative Care Unit underwent survival estimation by two independent physicians after a complete medical exam performed during the first day of admission. An independent research nurse also assessed each patient during the first day of admission. The assessment included activity, pain, nausea, depression, anxiety, anorexia, dry mouth, dyspnea, dysphagia, weight loss, and cognitive status. After the assessment was completed, patients were followed until discharge or death. In 47 evaluable patients, logistic regression showed a significant correlation between survival and dysphagia, cognitive failure, and weight loss. Accordingly, an "indicator of poor prognosis" was considered to exist in any patient who demonstrated weight loss of 10 kg or more plus cognitive failure (Mini-Mental State Questionnaire less than 24) plus dysphagia to solids or liquids. This indicator had a similar level of sensitivity, specificity, and overall accuracy, and a higher level of significance as compared with the assessment by physician #1 and physician #2, respectively. Our data suggest that three simple determinations, which may be performed by a nurse, can predict survival more or less than 4 wk as well as the assessments of two skilled physicians. These results need to be confirmed in other trials with large numbers of patients. Perhaps confirmation of these results and identification of other prognostic factors will result in staging systems for survival estimation of terminally ill cancer patients.