We provide evidence that the commercially available preparations of glycopeptides for intravenous application are well tolerated by endothelial cells when applied in concentrations less than 5 mg/ml. Since the antibiotics tested are administered at maximal concentrations of 10 mg/ml, the dose range used in our in vitro experiments (5 and 10 mg/ml) mimics possible clinical concentrations at the site of infusion. Similar concentrations may be reached by retrograde intravenous pressure infusion techniques (10-12). We have demonstrated that these high concentrations lead to considerable endothelial cell damage. These findings may explain the common side effect associated with intravenously applied glycopeptides namely pain and phlebitis at the site of infusion (2, 13). Figure 1 shows that a detrimental effect measurable after 20 min occurs only using vancomycin solutions at concentrations of 10 mg/ml, whereas already a dilution to 5 mg/ml renders the solutions more compatible to HUVEC. These data are in line with the observation that slow intravenous application of glycopeptides into large veins can largely prevent the occurrence of local phlebitis. Alternatively, the occurrence of phlebitis should be avoidable by diluting the manufacturers' preparations at least to 2-5 mg/ml and not 10 mg/ml as recommended by the manufacturer of vancomycin. The same aspects need to be considered for use of glycopeptides for retrograde high pressure infusion. The tolerance of intravenously applied antibiotics has previously been tested in animal models (4). Our model of human venous endothelial cells for testing antibiotic solutions for intravenous compatibility provides a valuable alternate model. In conclusion our data show that the commercial preparation of teicoplanin is more compatible for HUVEC than those of vancomycin.