The aim of this study was to determine the rate, risk factors and outcomes of catheter-related bloodstream infections (CRBSIs) in patients in a paediatric intensive care unit (PICU). A prospective cohort study was performed in King Abdulaziz Medical City, Riyadh, Saudi Arabia; a 650-bed academic/tertiary care centre with a combined 10-bed medical and surgical PICU. All patients admitted to the PICU from July 2000 to February 2003 who had a central line placed were monitored for the development of bloodstream infection (BSI) from insertion until 48 h after removal. Four hundred and forty-six patients with 2493 central-line-days were documented; 273 (55%) were male and the mean age was 2.6 years. Of the 446 patients, 278 (56%) had congenital heart disease, 108 (22%) had genetic disorders and/or congenital malformations, 55 (11%) had respiratory disease, and 42 (8%) had trauma. There were 50 episodes of CRBSI in 46 patients with a rate of 20.06 per 1,000 central-line-days and a device-utilization rate of 57%. Of these 50 episodes, 24 (48%) were polymicrobial, 16 (32%) were due to Gram-negative organisms, five (10%) were due to Gram-positive organisms, and five (10%) were fungal. The most common organisms isolated were Klebsiella pneumoniae (N=12, 16%), coagulase-negative staphylococci (N=10, 14%) and Pseudomonas aeruginosa (N=8, 11%). The mean duration of line insertion was 11.8 days for CRBSI patients and 4.22 days for non-BSI patients (P<0.0001). The mean PICU stay was 30.20 days for CRBSI patients and 6.35 days for non-BSI patients (P<0.0001). BSI occurred more often in catheters inserted in the PICU compared with the operating room, and in the femoral site compared with jugular or subclavian sites (P<0.001). In multiple logistic regression analysis of the risk factors, CRBSI patients were more likely to have multiple central lines [odds ratio (OR) 9.19; 95% confidence intervals (CI): 3.76-22.43), the line was more likely to be used for total parenteral nutrition (OR: 8.69; 95% CI: 3.5-21.4), and guidewire exchange was more likely to be performed on the line. CRBSI was not associated with a higher mortality rate. The CRBSI rate in our hospital is high compared with that reported by the National Nosocomial Infection Surveillance system. This study has established a benchmark for future comparisons. Additional studies from Saudi Arabia are necessary for national comparison and development of preventive measures.