Either fentanyl or Innovar (fentanyl, 0.05 mg/ml, and droperidol 2.5 mg/ml) was administered to supplement nitrous oxide anesthesia for operations on 29 patients. Both fentanyl and Innovar depressed the slope of the rebreathing CO2response curve during operation to 42 per cent ± 6 (mean of all intraoperative values, ± SE) of the awake control value. Following the last injection of drug but with continuation of operation, the slope increased such that it was 77 percent r S of control on the patients' arrival in the recovery room. The slope continued to increase to a peak of 103 per cent ± 5 of control. Soon thereafter respiratory depression recurred, as indicated by a decline in the slope to 55 per cent ± 5 of control, with a subsequent gradual return to 85 per cent ± 8 of control 230 minutes after the last injection. This biphasic response occurred in 90 per cent (26 of 29) of the patients treated either with fentanyl alone or with Innovar. Full recovery appeared to be more rapid with Innovar than with fentanyl alone. Droperidol did not augment and may have attenuated fentanyl-induced respiratory depression.