To investigate the epidemiologic and microbiological aspects of long-term central vein catheter (CVC)-associated bloodstream infections (CABSI) in children <18 years old treated at the hemato-oncology unit during 1998-2008. The two long-term access devices used were Hickman and Port-A-Cath catheters. Information retrieved included demographic data, baseline pathologies, methods of insertion, anatomical insertion sites, duration of use, microbiological, and antibiotic susceptibility data and outcome. There were 178 CABSI episodes; average number of episodes/1,000 catheter days was 4.7. More CABSI episodes were recorded among patients with Hickman catheter than in patients with Port-A-Cath catheter (5.05 vs. 3.57/1,000 catheter days, P = 0.059). The CVC was removed due to BSI in 52/178 (29.2%) episodes. Overall, 243 pathogens were isolated (144 Gram-negative, 92 Gram-positive, and 7 Candida spp). More Enterobacteriaceae spp. were isolated in CABSI in patients with Hickman catheters than in patients with Port-A-Cath catheters (35/103, 34%, vs. 10/65, 15%, P = 0.008); more coagulase-negative staphylococci were isolated in patients with Port-A-Cath catheters than in patients with Hickman catheters (25/65, 38.5%, vs. 23/103, 22.3%, P = 0.02). No differences in pathogen distribution were found between CABSI recorded for jugular versus subclavian veins, open versus close inserted-CVC or for CVC requiring removal versus those treated conservatively. No fatalities directly related to CABSI were recorded. CABSI rates were higher in patients with Hickman catheters compared with those with Port-A-Cath catheters; Gram-negative organisms were the dominant etiologic agents of CABSI; CABSI in patients with Hickman catheters had different etiologies compared with patients with Port-A-Cath catheters.