Intranasal cocaine is used frequently as a local anesthetic during many rhinolaryngologic procedures. Although its "recreational" use in high doses has been associated with chest pain and myocardial infarction, this association has not been established when cocaine is used in low doses as a topical anesthetic, and its effect on the coronary vasculature of humans is unknown. We studied the effects of intranasal cocaine (10 percent cocaine hydrochloride; 2 mg per kilogram of body weight) on the blood flow in and dimensions of the coronary arteries and on myocardial oxygen demand in 45 patients (34 men and 11 women, 36 to 67 years of age) who were undergoing cardiac catheterization for the evaluation of chest pain. Heart rate, arterial pressure, blood flow in the coronary sinus (measured by thermodilution), and the dimensions of the epicardial left coronary artery (measured by quantitative arteriography) were measured before and 15 minutes after the intranasal administration of saline (in 16 patients) or cocaine (in 29). No variables changed after the administration of saline. After cocaine was administered, the heart rate and arterial pressure rose, the coronary-sinus blood flow fell (from a mean [+/- SD] of 149 +/- 59 ml per minute to 124 +/- 53 ml per minute), and the diameter of the left coronary artery decreased by 8 to 12 percent (P less than 0.01 for all comparisons). No patient had chest pain or electrocardiographic evidence of myocardial ischemia after the administration of cocaine. Subsequently, the administration of the alpha-adrenergic blocking agent phentolamine caused all these values to return to base-line levels. There was no difference in response between the patients found to have disease of the left coronary artery (n = 28) and those without such disease (n = 17). We conclude that the intranasal administration of cocaine near the dose used for topical anesthesia causes vasoconstriction of the coronary arteries, with a decrease in the coronary blood flow, despite an increase in myocardial oxygen demand, and that these effects are mediated by alpha-adrenergic stimulation. It is reasonable to assume that these effects would be more pronounced at the much higher doses associated with the recreational use of cocaine.