The characteristics of the neuromuscular blockade produced by prolonged succinylcholine infusion were compared in 40 patients anesthetized with either nitrous-oxide-isoflurane (0.75-1.50% inspired) or nitrous-oxide-fentanyl. Neuromuscular transmission was monitored using train-of-four stimulation and the infusion rate was adjusted to keep the first twitch at 10-15% of its control value. Initially, all patients exhibited a depolarizing-type block, and the infusion rates were similar in the isoflurane (61 micrograms . kg-1 . min-1) and fentanyl (57 micrograms . kg-1 . min-1) groups. Tachyphylaxis developed in both groups and correlated well with the onset of non-depolarizing (phase II) block. Both occurred sooner and at a lower cumulative dose in the isoflurane groups. After 90 min, infusion rates were similar in both groups (isoflurane: 107 micrograms . kg-1 . min-1, fentanyl;: 93 micrograms. kg-1 . min-1). After the infusion was stopped, the recovery of the train-of-four ratio was inversely related to the dose and duration of exposure to succinylcholine, and was slower with nitrous-oxide-isoflurane anesthesia. After 10 min of recovery, patients receiving isoflurane exhibited train-of-four ratios of 0.5 or less after 8.5 mg/kg succinylcholine and 103 min. Corresponding figures for fentanyl patients were 13 mg/kg and 171 min. The block in all 13 patients (eight with isoflurane, five with fentanyl) who did not recover spontaneously was antagonized successfully with atropine and neostigmine. It was concluded that with succinylcholine infusion of 90 min or less, isoflurane accelerates the onset of tachyphylaxis and phase II neuromuscular block without affecting succinylcholine requirements. These results, with isoflurane, were similar to those reported previously with enflurane or halothane.