Fentanyl and alfentanil produce very similar electroencephalographic (EEG) changes in humans. With increasing serum concentrations of either narcotic, progressive slowing in frequency occurs. This narcotic effect on the brain was quantitated using off-line EEG power spectrum analysis. During EEG recording, six unpremedicated patients received a fentanyl infusion (150 micrograms/min), and six received alfentanil (1,500 micrograms/min) until a specific level of EEG depression (delta waves) occurred. Timed arterial blood samples were obtained for measurement of the narcotic serum concentrations. The narcotic-induced EEG changes were found to lag behind (in time) the serum narcotic concentration changes. To accurately relate EEG changes to serum narcotic concentrations, a pharmacodynamic model (inhibitory sigmoid Emax) was combined with a pharmacokinetic model that incorporated an "effect" compartment. (The effect compartment is the separate pharmacokinetic compartment where drug effect is directly proportional to drug concentration. It is the effect site.) The magnitude of the time lag was quantitated by the half-time of equilibration between serum narcotic concentrations and concentrations in the effect compartment. With fentanyl a significantly greater time lag was present (half-time = 6.4 +/- 1.3 min; mean +/- SD) than with alfentanil (half-time = 1.1 +/- 0.3 min). This difference in time lag between blood concentration and effect may be due to the larger brain-blood partition coefficient for fentanyl. The steady-state serum concentration that caused one-half of the maximal EEG slowing was 6.9 +/- 1.5 ng/ml for fentanyl, compared with 520 +/- 163 ng/ml for alfentanil.(ABSTRACT TRUNCATED AT 250 WORDS)