With the increasing use of artificial kidneys, numerous reports have appeared describing the pharmacokinetics of administered drugs in dialysis patients. Unfortunately, different investigators use different measures of dialysis clearance in reporting their results. Few studies have appeared in which actual measurements have been made in blood and dialysate as well as in plasma to experimentally show the variability of individual measurements and to demonstrate the inaccuracy of certain clearance measurements. We do so here, using the drug ethambutol. The effect of the artificial kidney on the removal of ethambutol was investigated in four uremic patients undergoing chronic hemadialysis. Ethambutol was administered by i.v. infusion over 30 min. Hemodialysis started at the end of drug infusion. Blood, plasma, and dialysate samples were collected periodically over 3 hr and analyzed for ethambutol content. Dialysis clearance was calculated by arterial-venous difference and by simultaneous dialysate measurement. The extraction efficiency of the hollow fiber dialyzers ranged from 36.2% to 43.8% in terms of blood and from 38.0% to 45.4% in terms of plasma. The mean clearance values due to dialysis were 108.08 and 88.1 ml/min measured with plasma and blood as body fluids of reference, respectively. Dialysis clearance calculated by dialysate measurement had a mean of 85.9 ml/min expressed as plasma and 74.7 ml/min expressed as blood. This study demonstrates that dialysis clearance when calculated using A-V difference and plasma flow is generally underestimated, particularly for a drug which extensively partitions into red blood cells. Ethambutol had a partition coefficient (blood/plasma) of greater than 1 in all four patients. The beta phase exhibited a mean half-life of approximately 2 hr on dialysis in comparison to off dialysis half-lives of 7 hr or longer in renal failure. Although ethambutol exhibits a markedly reduced half-life of the drug during hemodialysis, its recovery in the dialysis fluid during a 3-hr dialysis period constitutes only a small fraction of the dose administered.