Is pneumonia "the old man's friend"--a terminal event for patients who will otherwise die soon of underlying chronic disease? If so, chronological age might influence treatment policy. We investigated the predictors of 2-year mortality after patients' admission to hospital for community-acquired pneumonia, and focused on the predictive value of age. In a prospective cohort study 141 consecutive patients were admitted to hospital with community-acquired pneumonia. Clinical, laboratory, and sociodemographic data were collected on admission. Comorbidity was categorised as mild, moderate, or severe by a physician based on the patient's medical history. Survival was assessed at 24 months after discharge. 22 (16%) patients died in hospital. Of the remaining 119, 38 (32%) died over the next 24 months. In a Cox model, 2-year mortality was independently related to severe comorbidity (relative risk [RR] = 9.4) or moderate comorbidity (RR = 3.1), and to haematocrit less than 35% (RR = 2.9) (all p < or = to 0.005). However, compared with patients aged 18-44 years, patients aged 45-64 (RR = 0.84), 65-74 (RR = 1.28), and 75-92 (RR = 1.99) were not significantly more likely to die during the 24 months after discharge (all p > or = to 0.2). Old age should not be a sole criterion for withholding aggressive treatment of community-acquired pneumonia.