Tracheostomy continues to be a standard procedure for the management of long-term ventilator-dependent patients. Traditionally the procedure has been performed by surgeons in the operating theater using an open technique. This routine practice has recently been challenged by the introduction of bedside percutaneous dilatational tracheostomy (PDT), which has been reported to be a cost-effective alternative. The purpose of this study is to evaluate and compare the safety, procedure time, cost, and utilization of percutaneous and surgical tracheostomies at a university hospital. A retrospective medical chart review was performed on all ventilator-dependent intensive care unit patients at the University of Virginia Medical Center undergoing tracheostomy during a 23-month period beginning December 26, 1996. Of the 213 patients identified for review, 74 and 139 patients received percutaneous and surgical tracheostomies, respectively. Of 74 percutaneous tracheostomies, 73 reviewed were performed by general surgeons, pulmonary physicians, or anesthesiologists in the intensive care unit; all open tracheostomies were performed by surgeons in the operating room, and one percutaneous procedure was performed in the operating room. Perioperative complications occurred in five of 74 patients (6.76%) during PDT; of these, three patients (4.1%) experienced major complications requiring emergent operative exploration of the neck. Three patients (2.2%) experienced perioperative complications during surgical tracheostomy. The mean procedure time was significantly shorter for the percutaneous procedure. Average charges per patient in an uncomplicated case including professional fees, inventory, bronchoscopy (if performed), and operating room charges were $1753.01 and $2604.00 for percutaneous and standard tracheostomies, respectively. These charges do not include the charges associated with surgical intervention after PDT complications. In contrast to previously published reports showing complications clustered during a physician's first 30 percutaneous cases, our study demonstrated no relationship between complication occurrence and physician experience. That is, no learning curve associated with performing PDT was evident. In addition there was no association seen between physician specialty and complication rate. PDT in the intensive care unit costs less than surgical tracheostomy performed in the operating room and can be performed in less time. Several other studies have recommended that bronchoscopy during PDT provides additional safety; however, in our series all three major complications took place during bronchoscopy-assisted percutaneous procedures. Our series suggests that PDT carries an appreciable risk of major complications. Careful patient selection and additional experience with the procedure may decrease complication rates to an acceptable level.