To assess whether vancomycin administration at the time of central venous catheter insertion would prevent catheter-related sepsis (CRS) in immunocompromised patients, 98 cancer patients were entered into a randomized placebo-controlled trial. Patients were stratified according to whether they were having therapy for acute leukaemia or undergoing bone-marrow transplantation (group A) or required chemotherapy for a solid tumour (group B). Eligible patients were randomized to receive either 500 mg vancomycin in 250 ml of 0.9% saline prior to catheter insertion followed by a further 500 mg infused via the established catheter, or the same regimen with 0.9% saline alone. In group A, there were 32 instances of CRS occurring in 20 of the 35 evaluable catheters (57%). Six catheters were removed because of CRS. No significant difference was found in the incidence of CRS between the two arms. Of the 37 evaluable catheters in group B, CRS occurred in 6 (16%), and none of the catheters required removal because of CRS. Again, no significant differences were found in the incidence of CRS between the patients given vancomycin or placebo. These findings indicate that the incidence of CRS is greater in patients who have more severe and prolonged immunosuppression and that vancomycin prophylaxis fails to reduce CRS in patients undergoing chemotherapy for malignant disease.