Immediately following the connection of pediatric patients to cardiopulmonary bypass we have consistently observed a steep decrease in fentanyl plasma concentration (74 +/- 8.7%) (mean +/- SD), much greater than would have been expected from hemodilution alone (50.6% +/- 12.0%) (p less than 0.0001). Priming of the pump with 20 ng/ml of fentanyl before connection to the patients did not prevent this phenomenon. In order to study the possibility that fentanyl is sequestered by the bypass, levels of the primed drug in the bypass were assessed before connecting the pump to the children and a steep fall from 20 ng/ml to zero was shown before initiation of bypass. Pharmacokinetic assessment of fentanyl in a closed pump circuit showed that levels of 120 ng/ml fall to 2 ng/ml within 3 min and remain stable at the lower concentration for at least 30 min. Further studies have identified the membrane oxygenator as the major site of fentanyl sequestration. Concentrations across the membrane fall from 120 ng/ml to 10 ng/ml. The attached siliconized tubing is associated with a minor binding effect sufficient to reduce concentrations from 110 to 84 ng/ml. The pvc tubing, aluminium heat exchanger and plastic reservoir had no binding effect on fentanyl. The possibility that a decrease in fentanyl protein binding caused the fall in serum concentration was checked in 5 patients undergoing open heart surgery. After initiation of the cardiopulmonary bypass, there was a significant decrease in albumin serum concentrations from 32.0 +/- 2.3 mM to 15.0 +/- 1.6 mM (p less than 0.0001).(ABSTRACT TRUNCATED AT 250 WORDS)