The tendency of highly hypnotizable participants to bias their retrospective perceptual reports in response to instructional demands was reexamined with the addition of low-hypnotizable control participants instructed to simulate hypnosis. Mean scores of high-hypnotizable participants and simulators did not differ, but the responses of simulators to the demand instruction was less variable than those of high-hypnotizable participants, and the shape of the response distribution was different. Unlike simulators, some high-hypnotizable participants who had reported changes in perception that were consistent with a hypnotic suggestion subsequently reported changes opposite to those suggested by a demand instruction. These data were interpreted as suggesting that the responses of high-hypnotizable participants to both the demand instruction and the preceding hypnotic suggestion were not entirely due to compliance.