Medical treatment is usually accompanied by a warning about potential side effects. While constituting an important component of informed consent, these warnings may themselves contribute to side effects via the placebo effect. We tested this possibility using a 2 × 2 between-subjects design. Under the guise of a trial of a new hypnotic, 91 undergraduates experiencing difficulty sleeping were allocated to receive a warning about a target side effect (either increase or decrease in appetite, counterbalanced) or no warning and then to receive placebo treatment or no treatment for one week. Placebo treatment led to significantly better sleep on almost all self-reported outcomes, suggesting a placebo effect for reported sleep difficulty. Actigraphy recordings were unaffected by treatment. There was a clear effect of the warning in that placebo treated participants who were warned about side effects were much more likely to report the target side effect than those not warned about side effects. Implications for clinical practice are discussed.