Until now, studies confirming the safety of glycopeptide restriction in the empirical treatment of prolonged fever and neutropenia included only nine children. In an open-label observational study, the use of teicoplanin in paediatric oncology patients was investigated. A period of unrestricted use (2001-2003) was compared with a second period (2004) following implementation of a restrictive treatment guideline. Empirical first-line treatment consisted of piperacillin/tazobactam; in 2004, fosfomycin was added after 72 h as the second-line combination instead of teicoplanin. In total, 213 episodes (n=163 in 2001-2003; n=50 in 2004) managed with teicoplanin or fosfomycin (only 2004) were eligible. Empirical treatment of fever of unknown origin with teicoplanin was reduced by 97%. In 2004, the mean length of stay was 0.4 days shorter, no infection-related death occurred and no vancomycin-resistant enterococci were detected. Restriction of empirical glycopeptides is safe in paediatric cancer patients after first-line treatment with piperacillin/tazobactam. Fosfomycin appears to offer a feasible and cost-saving alternative in second-line combination therapy.