The aim of this study was to examine the relationships among classical conditioning, expectancy, and fear in placebo analgesia and nocebo hyperalgesia. A total of 42 healthy volunteers were randomly assigned to three groups: placebo, nocebo, and control. They received 96 electrical stimuli, preceded by either orange or blue lights. A hidden conditioning procedure, in which participants were not informed about the meaning of coloured lights, was performed in the placebo and nocebo groups. Light of one colour was paired with pain stimuli of moderate intensity (control stimuli), and light of the other colour was paired with either nonpainful stimuli (in the placebo group) or painful stimuli of high intensity (in the nocebo group). In the control group, both colour lights were followed by control stimuli of moderate intensity without any conditioning procedure. Participants rated pain intensity, expectancy of pain intensity, and fear. In the testing phase, when both of the coloured lights were followed by identical moderate pain stimuli, we found a significant analgesic effect in the placebo group, and a significant hyperalgesic effect in the nocebo group. Neither expectancy nor fear ratings predicted placebo analgesia or nocebo hyperalgesia. It appears that a hidden conditioning procedure, without any explicit verbal suggestions, elicits placebo and nocebo effects, however we found no evidence that these effects are predicted by either expectancy or fear. These results suggest that classical conditioning may be a distinct mechanism for placebo and nocebo effects.