Perfused biofilms of Enterococcus faecalis reached a pseudo-steady state, with respect to bioburden and the numbers of cells liberated within the perfusates. Biofilm growth rate was slow (approximately 0.006 doublings/h) relative to batch culture (approximately 0.01 doublings/h). MICs were determined for vancomycin (0.12 mg/L) and teicoplanin (1.2 mg/L) in batch culture. Steady-state biofilms (24 h and 48 h) were perfused continuously for 96 h with medium containing antibiotic at 4 X MIC. Susceptibility was assessed as cfu in the perfusate, indicating growth inhibition, and as cfu for the parent biofilm. Vancomycin at these levels had little or no effect on either parameter, whilst teicoplanin produced a temporary (30 h) reduction in growth rate (99.99% for 24 h biofilms, 50% for 48 h biofilms). Antibiotic concentrations were raised to therapeutic (trough) levels (vancomycin, 5 mg/L; teicoplanin, 12 mg/L) and applied continuously to 24 h old biofilms. Neither agent affected viability of the biofilm over 96 h. Biofilm growth rate, however, was decreased markedly over the first 8-10 h of antibiotic treatment and was maintained at the reduced level for approximately 40 h. Thereafter growth of the biofilms gradually returned to pre-exposure levels. The pattern of recovery was different for the two agents suggesting that different mechanisms might be involved. Accordingly biofilms were exposed to successive and concurrent 24 h treatments with vancomycin (5 mg/L) and teicoplanin (12 mg/L). This led not only to a further 2-3 log reduction in the growth rate but also to a 3 log reduction in the viability of the parent biofilm. Such synergy between the glycopeptide agents might have therapeutic implications for the treatment of E. faecalis infection.