All 5,047 consecutive inpatients admitted to the Internal Medicine Division of a teaching hospital (Zieglerspital, Berne) between 1982 and 1985 were registered in accordance with the CHDM (Comprehensive Hospital Drug Monitoring) questionnaire of adverse drug reactions (ADRs). Of them, 2,439 were treated with at least one potassium losing diuretic. The hospital records of the patients were reviewed with particular regard to serum potassium levels, and on the basis of this evaluation, the patients were assigned to four different diuretic treatment groups, and the incidence of hypokalaemia related to diuretic treatment was estimated. The overall rate of occurrence of hypokalaemia was 21.1% at a serum potassium level < 3.5 mmol.l-1, and 3.8% < 3.0 mmol.l-1. Hypokalaemia of less than 3.5 mml.l-1 developed 24.9% (217/870) of patients treated with potassium losing diuretics alone; in 19.7% (101/513) treated with potassium losing diuretics in conjunction with potassium substitution, in 15.1% (66/438) treated with a combination of diuretics (potassium losing with potassium sparing), and in 20.0% (12/60) treated with combined diuretics and potassium substitution. Only the differences between the first and the two subsequent groups were statistically significant. The overall incidence of hypokalaemia below 3.0 + mmol.l-1 was significantly lower in the patients on combined diuretics without potassium substitution than in the patients on potassium losing diuretics with potassium substitution. Oral or parenteral administration of glucocorticoids (prednisone 5 to 2,000 mg/d) was a significant risk factor for hypokalaemic events. beta 2-Adrenoceptor agonists had not effect. The patient's age, sex, renal function and numbers of drugs received were evaluated in a multivariate analysis, in order to take into account their influence on the risk of developing hypokalaemia. The number of drugs above 12 (and, less importantly, female sex) was the main risk factor for this ADR. The comparison between hypokalaemia and hyperkalaemia in this group of inpatients showed the significance of reduced renal function in the occurrence of hyperkalaemia.