Animal research has indicated that the activity of the hypothalamic-pituitary-adrenal axis can be influenced by classical (Pavlovian) conditioning procedures. To test the hypothesis that alterations in plasma cortisol levels can be conditioned in humans, the present study used a prospective, randomized, double-blind, placebo-controlled design in which a distinctively flavored beverage was paired with p.o. administration of dexamethasone. Twenty-five healthy men were randomly assigned to one of two groups. During the conditioning phase of the study, subjects in the experimental group received three conditioning trials (pairings of a distinctively flavored beverage with a capsule containing 5 mg dexamethasone) separated by 1 wk recovery periods. Subjects in the control group were treated identically, except that the capsule contained a placebo. During the test phase, all subjects underwent a test day (reexposure to the distinctively flavored beverage before receiving a placebo capsule) and a comparison day (no exposure to the beverage or the capsule). Plasma cortisol was assessed repetitively before and after administration of the beverage and capsule, as were possible confounding factors, including: behavioral variables, psychological distress, aversive reactions to the beverage, and expectations of treatment. After reexposure to the beverage and administration of a placebo capsule (conditioned stimuli), the experimental group had significantly higher levels of plasma cortisol than the control group, after controlling for variability in baseline levels of cortisol (F(5,60)=3.09; P=0.015) that could not be explained by differences in other study variables. No differences in cortisol levels were found on the comparison day. These results support the study hypothesis that changes in plasma cortisol levels can be classically conditioned in humans by pairing a distinctive beverage with p.o. administration of dexamethasone.