Potassium homeostasis is essential for normal myocardial function, and low serum potassium may cause fatal arrhythmias. However, the association of low potassium and long-term mortality and morbidity in heart failure (HF) is largely unknown. We studied 6845 HF patients in the Digitalis Investigation Group trial with serum potassium levels < or =5.5 mEq/L. Of these, 1189 had low potassium (<4 mEq/L). Propensity scores for low potassium were calculated for each patient and were used to match 1187 low-potassium patients with 1187 normal-potassium (4-5.5 mEq/L) patients. Effects of low potassium on outcomes were assessed using matched Cox regression analyses. All-cause mortality occurred in 379 (rate, 1103/10 000 person-years) normal-potassium and 441 (rate, 1330/10 000 person-years) low-potassium patients, respectively, during 3437 and 3315 years of follow-up [hazard ratio (HR), 1.25; 95% confidence interval (CI), 1.07-1.46; P = 0.006]. Cardiovascular mortality occurred in 297 (864/10 000 person-years) normal-potassium and 356 (1074/10 000 person-years) low-potassium patients (HR, 1.27; 95% CI, 1.06-1.51; P = 0.009). Cardiovascular hospitalization occurred in 610 (rate, 2553/10 000 person-years) normal-potassium and 637 (rate, 2855/10 000 person-years) low-potassium patients (HR, 1.13; 95% CI, 0.99-1.29; P = 0.082). In a cohort of ambulatory chronic systolic and diastolic HF patients who were balanced in all measured baseline covariates, serum potassium <4 mEq/L was associated with increased mortality, with a trend towards increased hospitalization.