Classical conditioning of skin conductance responses was studied in 26 healthy volunteers and 31 patients who received general anesthesia (0.5 MAC isoflurane-70% nitrous oxide) for minor gynecologic surgery to determine whether there was autonomic evidence of new learning during anesthesia. In the conditioning phase, a loud noise was paired with a target word (conditioned stimulus [CS]). Conditioning was established if the CS produced conditioned responses, i.e., greater skin conductance responses than other, "filler words." In the subsequent elicitation phase, we investigated whether it was possible to elicit similar conditioned responses to the CS when it was presented without noise about 3 h later. Conditioning was established in the control group, who received no anesthesia. There were differences in response magnitudes of 0.32 and 0.27 mu siemen between responses to the CS and filler words, P less than 0.001. In contrast, there was no evidence that conditioning was established during anesthesia. In the elicitation phase, conditioned responses to the CS could be elicited in volunteers; i.e., response magnitudes were greater for the CS than other words (during the first two blocks of trials), with differences of 0.19 to 0.48 mu siemen, P less than 0.05 to P less than 0.001. In contrast, responses to the CS could not be elicited in patients after anesthesia. There was also no evidence of recognition of words that were presented during anesthesia in patients, but recognition was manifested in volunteers. Thus, in contrast to our previous demonstration of conditioning in subjects who received 30% nitrous oxide in oxygen, there was no evidence of conditioning in anesthetized patients.