Discussion We compared 2 similar groups of ICU patients admitted who had normal serum levels of potassium and creatinine at the time of admission. The patients had a variety of medical and surgical diagnoses. Compared with the control group, patients who were treated empirically with a potassium supplement added to maintenance intravenous fluid at a rate of 72 to 144 mmol/d received significantly fewer potassium boluses throughout their ICU stay. This difference between the 2 groups was true for all diagnostic categories. Receiving fewer potassium boluses did not affect the ICU length of stay, but it did decrease the number of invasive procedures and the risk associated with potassium administration. In addition, we calculated a direct savings in material costs of approximately $231 per patient. Most important, because a potassium bolus is administered in response to low serum levels of potassium, the patients in the treatment group had significantly fewer episodes of hypokalemia. Limitations of our study include the lack of random assignment of patients and data collection for a single site. However, because of the physical and financial benefit to patients and the decrease in workload for health care providers, preemptive administration of potassium to prevent episodes of hypokalemia should be considered for patients admitted to critical care areas who require intravenous fluid at a rate of 75 mL/h or greater. Further studies with a prospective design and random assignment to allow for the generalizability of the results are needed. Such studies may indicate a need to