A prospective study was done in 139 intravascular catheters (IVCs) that had been removed for different reasons. The purpose of the study was to compare laboratory procedures for the diagnosis of catheter-related infections and also to attempt to clarify the present controversy regarding the portal of entry of such infections. The IVCs were removed by one of us and multiple samples were studied according to a standard procedure. Semi-quantitative cultures were performed of the tips, the interior of the hub, and the skin around the insertion point. Quantitative cultures were performed of the infusion fluid and of the IVC tips. Of the 139 IVCs studied, 53 (38.1%) were infected (greater than or equal to 15 colony-forming units per plate in the semiquantitative culture). Semi-quantitative and quantitative cultures gave comparable results, but the semiquantitative procedure proved to be easier and faster. All but three infected catheters had a positive (greater than or equal to 15 colony-forming units per plate) skin and/or hub culture (superficial cultures), with microorganisms identical to those isolated in the IVC tip. Our results showed two possible and differentiable portals of entry. Thirty (56.6%) had external origin (semiquantitative skin culture positive), 12 (22.6%) had an internal origin (semiquantitative hub culture positive), and 8 (15.1%) had both origins. All catheters with negative superficial cultures had a negative tip. The predictive value of positive superficial cultures in the diagnosis of catheter-related infection was 66.2% and that of negative cultures was 96.7%. In patients with suspected catheter-related infections but negative superficial cultures, the possibility of infection may reasonably be ruled out, thereby avoiding many unnecessary catheter withdrawals.