Procedures of classical conditioning in animals and man have provided evidence that most psychophysiological responses can be acquired by repeated association with previous neutral stimuli. Many animal studies reported on conditioned changes in the blood glucose level; the nature of the conditioned response (CR), hypo- or hyperglycemia, however, seems to vary with experimental procedures. The present study aimed to elicit conditioned blood glucose changes in human subjects. Thirty male volunteers participated in five sessions each. The sessions were separated by 3 days, with identical time course and procedure. The subjects were informed that we wanted to test the effects of insulin or placebo injections on cognitive functioning, and were kept busy with pseudotests. In four sessions, subjects were injected with 0.035 IU/kg body weight of human insulin as the unconditioned stimulus (US), which induced the expected fall in blood glucose level below 50 mg/dl (UR). Injections were accompanied by a specific stimulus compound (conditioned stimulus, CS) in half of the subjects. In the fifth session, the CS was associated with a placebo injection. About half of the subjects showed a change from the baseline level of blood glucose of more than 10 mg/dl, which we would interpret as a conditioned response. Conditioning occurred more often in those subjects who were given a CS compound in addition to the injection, which itself together with the experimental environment may have been a sufficient CS.