To evaluate risk factors for infections associated with indwelling arterial catheters, 186 catheters were randomly allocated for either femoral or radial insertion in 155 critically ill patients. Femoral catheters were easier to insert and it was easier to obtain blood specimens from them. Rates of local infection at the insertion sites and rates of positive catheter-tip cultures were similar for femoral and radial catheters. Evidence of local infection was not predictive of a positive catheter-tip culture. Percutaneously inserted femoral and radial artery catheters had a similarly low incidence of catheter-associated infections. There was only one catheter-related infection, and no cultured catheter was judged the cause of bacteremia. Routine prophylactic replacement of arterial catheter systems may be unnecessary in critical care units where rates of arterial catheter-associated infections are low.