We studied the infectious risk of different methods of managing vascular catheters during long-term use. Consecutive surgical ICU patients requiring triple lumen catheters, pulmonary artery catheters, or arterial catheters for greater than 7 days were prospectively randomized to one of three management groups: a) percutaneous (PERC) puncture with every 7-day catheter change at a new site, b) no weekly change (NWC) with a new site when changed, or c) guidewire exchange (GWX) with every 7-day catheter change at the same site. In all groups, a catheter change was mandatory for a positive blood culture, skin site infection, or sepsis without a likely source. Cultures were obtained when clinically indicated and at the time of every catheter change. Catheter-related sepsis (CRS) was defined as a positive blood culture and catheter culture with the same organism. A total of 112 patients met evaluation criteria. There were no intergroup differences in age, primary diagnosis, severity of injury or illness, number of study days, number of protocol violations, route of catheterization, number of catheters present/patient day, catheter sepsis rate, or bacteremia rate. The NWC group demonstrated an increased number of days/catheter, fewer catheter/subcutaneous tract segment cultures/patient, and a reduced incidence of catheter tip colonization. These results occurred in a setting where the number of CRS episodes/patient was 0.17 for GWX, 0.22 for PERC, and 0.16 for NWC. We conclude that there is no difference in infectious risk between these three methods of long-term catheter management. The method with the least complications and expense should be used.(ABSTRACT TRUNCATED AT 250 WORDS)