59 patients with suspected central venous catheter related bacteraemia (CRB), while receiving parenteral nutrition, were studied prospectively. 41 (Group 1) were managed conservatively: cultures were taken from the catheter hub lumen, skin at the catheter entry site and peripheral blood; the catheter was then heparinised and locked for 24-48 h. The catheter was withdrawn only if cultures were positive; otherwise parenteral nutrition was resumed. In 18 patients the catheter was immediately withdrawn (Group 2) and the same cultures plus tip culture were performed. 13 patients of Group 1 with positive hub or skin cultures, had their catheters removed and 12 had blood cultures matching for the same micro-organism. Negative skin and hub cultures had a negative predictive value for CRB of 96%. A positive hub culture had a 100% positive predictive value for CRB. CRB was diagnosed in 11 patients out of the 18 in whom catheters were withdrawn immediately (Group 2). Thus, 1 out of 41 catheters and 7 out of 18 catheters were removed unnecessarily in Groups 1 and 2 respectively (p = 0.001, Fisher's test). In all, 24 CRBs were documented and 15 were due to coagulase negative staphylococci. The catheter hub was the commonest origin of CRB followed by the infusate and the skin. In febrile patients on parenteral nutrition, negative skin and hub cultures accurately predict or rule out CRB and should be used more often to avoid withdrawal of sterile catheters. © 1993.