To study the epidemiology -especially the impact of contaminated stopcocks -on central venous catheter (CVC) infection and catheter-related sepsis (CRS), semiquantitative (SQ) and quantitative (Q) culture methods and typing of coagulase-negative staphylococci (CNS) were employed in 49 neonates with clinical signs of sepsis while receiving parenteral nutrition in the paediatric intensive care unit. The patients were divided into two groups according to stopcock contamination: group A consisted of 18 patients (36%) with contaminated stopcocks and group B consisted of 31 patients (64%) with sterile stopcocks. Five specimens were obtained from each patient, in addition to that from the stopcock: a swab taken from the skin surrounding the catheter puncture site; the CVC tip; the intradermal segment (IDC); and samples of parenteral fluid and blood. A total of 294 specimens (392 sites) was cultured and micro-organisms were identified. All CNS isolated were typed by biotyping, antibiogram, plasmid analysis and pulsed-field gel electrophoresis (PFGE), and the discriminatory power of the typing methods was compared. The CVC tips were infected in 25 patients (51%); 15 (83%) in group A and 10 (32%) in group B. Sepsis was detected in 24 neonates (@YO), 13 in group A and 11 in group B. This was catheter-related in 15 patients (63%), 12 in group A and 3 in group B. CNS were recovered from 13 (52%) of 25 infected CVCs, nine in group A and four in group B. Sixty-five CNS isolates were recovered from these patients and belonged to 14 biotypes, 22 antibiograms, 22 plasmid profiles and 26 PFGE types. Typing showed that in six of nine patients in group A, CNS of the same type were recovered from the catheter tip and the stopcock, in one patient the catheter tip and skin isolates were the same and in two others the catheter tip isolates were different from stopcock and skin isolates. In all four patients in group B, different CNS types were recovered from CVC tips and skin. Bacteraemia was caused by CNS in 14 patients (58%), six in group A and eight in group B. v p i n g confirmed that nine cases (six in group A and three in group B) were catheter-related but five were not. SQ and Q culture methods and typing, especially by PFGE, allowed the study to determine that bacteria from contaminated stopcocks were frequently the source of CVC infection and CRS.