Context: Although potassium is critical for normal electrophysiology, the association between abnormal preoperative serum potassium level and perioperative adverse events such as arrhythmias has not been examined rigorously. Objective: To determine the prevalence of abnormal preoperative serum potassium levels and whether such abnormal levels are associated with adverse perioperative events. Design and Setting: Prospective, observational, case-control study of data collected from 24 diverse US medical centers in a 2-year period from September 1, 1991 to September 1, 1993. Patients: A total of 2402 patients (mean [SD] age, 65.1 [10.3] years; 24% female) undergoing elective coronary artery bypass grafting who were not enrolled in another protocol. The study population was identified using systematic sampling of every nth patient, in which n was based on expected total number of procedures at that center during the study period. Main Outcome Measures: Intraoperative and postoperative arrhythmias, the need for cardiopulmonary resuscitation (CPR), cardiac death, and death due to any cause prior to discharge, by preoperative serum potassium level. Results: Perioperative arrhythmias occurred in 1290 (53.7%) of 2402 patients, with 238 patients (10.7%) having intraoperative arrhythmias, 329 (13.7%) having postoperative nonatrial arrhythmias, and 865 (36%) having postoperative atrial flutter or fibrillation. The incidence of adverse outcomes was 3.6% for death, 2.0% for cardiac death, and 3.5% for CPR. Serum potassium level less than 3.5 mmol/L was a predictor of serious perioperative arrhythmia (odds ratio [OR], 2.2; 95% confidence interval [CI], 1.2-4.0), intraoperative arrhythmia (OR, 2.0; 95% CI, 1.0-3.6), and postoperative atrial fibrillation/flutter (OR, 1.7; 95% CI, 1.0-2.7), and these relationships were unchanged after adjusting for confounders. The significant univariate association between increased need for CPR and serum potassium level less than 3.3 mmol/L (OR, 3.3; 95% CI, 1.2- 9.5) and greater than 5.2 mmol/L (OR, 3.0; 95% CI, 1.1-8.7) became nonsignificant after adjusting for confounders. Conclusions: Perioperative arrhythmia and the need for CPR increased as preoperative serum potassium level decreased below 3.5 mmol/L. Although interventional trials are required to determine whether preoperative intervention mitigates these adverse associations, preoperative repletion is low cost and low risk, and our data suggest that screening and repletion be considered in patients scheduled for cardiac surgery.