The aim of this study was to analyse the diagnostic, empirical and therapeutic strategies adopted when a blood culture from a hospitalized child with a central venous catheter is 'positive', and to assess whether practices complied with the consensus adopted in our hospital, inspired by published recommendations. One hundred and ten cases of bacteraemia were studied prospectively. Investigations to determine whether the catheter was the cause of infection were carried out in 45% of cases, and the catheter was removed as recommended in 39% of cases. Of the patients that received empirical treatment, 56% received broad-spectrum antibiotics with no apparent clinical justification. Following susceptibility testing on the isolated strain, the antibiotic treatment was considered to be appropriate in 58% of cases. Overall, compliance with the consensus recommendations was poor. This was partly due to the high turnover rate of antibiotic prescribers.