Five hundred two central venous catheters inserted in 366 patients were evaluated prospectively over a one-year period to determine the frequency and risk factors associated with catheter-related sepsis. For study purposes, in cases in which catheter infection was suspected but the initial blood cultures were negative, the catheters were replaced by guidewire technique; otherwise, the catheters were routinely changed after 21 days by guidewire technique. A catheter-related infection was suspected in 190 cases (190/502, 38%). A diagnosis of catheter-related sepsis was established in 50 patients, which represents 10% of the total number of lines (502). Over a total of 6428 days of catheter use, the infection rate was 0.8 cases of sepsis per 100 catheter-days. Staphylococcus epidermidis, Staphylococcus aureus, and Candida spp. were the most frequently isolated aetiological agents of sepsis. On univariate analysis, six variables affecting the rate of catheter-related sepsis were identified: neutropenia for more than eight days (p < 0.001); AIDS (p < 0.001); haematological malignancy (p < 0.001); administration of total parenteral nutrition (p = 0.001); duration of site use (p = 0.04); and high APACHE II score (p = 0.04). The logistic regression analysis revealed that AIDS and haematological malignancies were independent risk factors of catheter-related sepsis. Catheter replacement over a guidewire was no more likely to be associated with sepsis than was percutaneous catheter insertion. In conclusion, although the incidence of established catheter infection is much lower than the incidence of suspected infection, in most cases of suspected infection it is wise to change the catheter with the guidewire technique and wait for culture of the tip, rather than to remove the catheter immediately. Such a policy may help reduce the number of unnecessary catheter removals.