Little is known about vascular access infections in patients with acute renal failure. We prospectively compared infection rates of dialysis catheters (DCs) and central venous catheters (CVCs) in patients in the intensive care unit treated with renal replacement therapy for acute renal failure. The same insertion and maintenance procedures were used for CVCs and DCs. To circumvent the allocation bias caused by severity of patient condition, only patients with both types of catheters were included. A total of 150 CVCs and 130 DCs were analyzed in 99 patients with a mean Simplified Acute Physiology Score II of 67 +/- 21. The major cause of acute renal failure was sepsis (62%). Hospital mortality was 62%. Mean catheter duration was shorter for DCs (6.7 +/- 4.4 days) than CVCs (7.8 +/- 4.9 days; P = 0.03). There was no difference between CVCs and DCs in cumulative incidence of catheter colonization (quantitative catheter cultures > or = 10(3) colony-forming units/mL; 4.7% versus 6.2%; P = 0.58) or incidence density of catheter colonization per 1,000 catheter days (5.9 versus 9.1; P = 0.44, respectively). There also was no difference between CVCs and DCs in cumulative incidence and incidence density regardless of whether catheters were placed at the internal jugular (P = 0.34 and P = 0.23) or femoral site (P = 0.57 and P = 0.80), respectively. Three cases of CVC-related bacteremia (the same microorganism responsible for both catheter colonization and blood culture result) were recorded, but none with DC use. When severity of patient condition is controlled for, epidemiological characteristics of colonization in CVCs and DCs are similar if similar infection control measures are used for insertion and maintenance.