Reports of late-onset cardiovascular complications following tricyclic antidepressant (TCA) overdose have led to a very conservative approach to these patients. Many patients have been hospitalized for continuous cardiac monitoring, regardless of the clinical presentation. Management algorithms based on clinical predictors of outcome have recently been proposed. We used the algorithm developed by Tokarski and Young to retrospectively evaluate the care of 33 TCA overdose patients admitted to our hospital over a 3-year period. We then identified 11 patients who could have been treated on an outpatient basis had the algorithm been employed. Ten were admitted to a monitored unit and spent a mean of 31.6 +/- 15.64 hours on the unit. None of the 11 patients developed complications during their hospital stay. Use of the algorithm would have resulted in an estimated cost savings of 13 hospital days and $14,000.