The aim of the study was to quantitatively predict the clearance of three antibiotics, amikacin, vancomycin, and teicoplanin, during continuous hemodiafiltration (CHDF) and to propose their optimal dosage in patients receiving CHDF. For this goal, in vitro CHDF experiments with a polyacrylonitrile (PAN) membrane were first performed using these antibiotics, and then the clearances were compared with in vivo CHDF situations determined in 16 critically ill patients. The in vitro CHDF clearances were described as the product of the outflow rate of a drain (Q outflow ) and the drug unbound fraction in artificial plasma, indicating that drug adsorption to the PAN membrane has minor effect on drug clearance in our settings. The observed in vivo clearances also agreed very well with the predicted values, with a product of Q outflow and plasma unbound fraction, when residual creatinine clearance (CL CR ) was taken into account (within a range of 0.67-to 1.5-fold for 15 of 16 patients). Based on these results, a nomogram of the optimized dosages of amikacin, vancomycin, and teicoplanin was proposed, and it was evident that Q outflow and residual CL CR are major determinants of the dosage and dosing interval for these antibiotics. Although the applicability needs to be confirmed with another type of membrane or higher Q outflow , our nomogram can help determine the dosage setting in critically ill patients receiving CHDF.