We prospectively studied the risk of catheter-related sepsis (CRS) in 75 critically ill patients who received total parenteral nutrition (TPN) through 158 pulmonary artery catheters (PACs) and 214 triple-lumen catheters (TLCs). We relied on semiquantitative cultures of the catheter tips, peripheral blood cultures in febrile patients and clinical response to catheter removal to diagnose catheter-related sepsis. The infection rate was 2.5% (4/158) of PACs and 6.5% (14/214) of TLCs (p = 0.124). Colonization rates were 29.1% for PACs and 32% for TLCs. PACs were left in place a significantly shorter length of time than TLCs, 3.1 vs 5.1 days (p less than 0.005). Guidewire exchanges and subclavian vein insertions were associated with a decreased rate of CRS when compared to new insertions and internal jugular vein insertions, respectively. We conclude that pulmonary artery catheters can be used safely for the delivery of hyperalimentation in critically ill patients with no increased risk for catheter-related sepsis compared to triple-lumen catheters. The use of the PAC in this manner allows for the use of a single central venous catheter for the delivery of hyperalimentation and hemodynamic monitoring.