In a prospective study, 10 patients (ASA II-IV) with different ailments received varying amounts of an alfentanil/midazolam mixture (consisting of 15 mg of alfentanil and 37.5 mg of midazolam) for analgesia and sedation in a typical ICU setting over a period of 3-14 days. Special interest focussed on potential residual activity of both drugs after termination of the treatment. Since both compounds interact with receptor sites in the CNS, neurophysiological measurements were performed before, during and after the sedation. Thus, EEG power spectra were used to evaluate the central effects of midazolam, while the late peak (greater than 50 ms) of the somatosensory-evoked potential (SEP) was used to determine the inhibitory effect of alfentanil on the propagation of efferent nerve volleys along the sensory tract. A residual activity of midazolam after long-term administration seems likely as the return in the fast beta-domain was slow, and the high power in the slow EEG frequency bands delta and theta still persisted 24 hours after treatment. However, rapid recovery of the amplitude of the late N100-peak in the evoked potential suggests no potential "overhang" of alfentanil even if administered for several days together with a benzodiazepine. Among the variables most suitable for the determination of a sufficient sedation, the spectral edge activity (the highest frequency component in the EEG incorporating 95% of the entire power spectrum) showed a close relationship to the applied dose. All patients except one, a former alcoholic with CT-verified widening of the cerebral ventricles, could be sufficiently sedated with alfentanil/midazolam.(ABSTRACT TRUNCATED AT 250 WORDS)