The mechanism of tolerance to nicotinic acid flushing was determined in subjects during a 5-day course of treatment. Objective measures of skin blood flow were used to confirm the development of tolerance. Plasma levels of nicotinic acid showed marked intraindividual variability but were not decreased with the development of tolerance. However plasma levels of 9-alpha 11-beta prostaglandin F2, a stable metabolite of prostaglandin D2, became undetectable in most subjects with the development of tolerance. Thus tolerance is not associated with decreased levels of nicotinic acid or development of tolerance to the prostaglandin mediator, but with decreased levels of the mediator.