The frequency of hyponatraemia varies from less than 1% to more than 40% in series reported from general hospitals. We performed a prospective study to evaluate its incidence and to determine the types of hyponatraemia in a dedicated cancer hospital. All patients admitted to the Department of Medicine were prospectively studied over 11 months. Hyponatraemia was defined as a serum Na level < or =130 mEq/l. Urine and blood samples were collected at baseline, and all consecutive hyponatraemic episodes were studied. One hundred and six patients developed 123 episodes of hyponatraemia. The observed incidence of hyponatraemia was 3.7%. Sodium depletion and syndrome of inappropriate antidiuretic hormone secretion (SIADH) each accounted for almost one third of all aetiologies. Serum urea and uric acid, urinary Na and fractional excretions of Na and urea were most useful for the differential diagnosis. The percentage of deaths observed in the hyponatraemic group, 19.5%, was higher than in the whole cancer population (6.3%) although no death was apparently due directly to hyponatraemia. Hyponatraemia is regularly diagnosed in cancer patients, but it was related to SIADH in only about one third of the cases. A higher mortality was observed in hyponatraemic patients than in normonatraemic patients.