Objectives.—To investigate the prophylactic and acute effects of autogenic training (AT) during a nitroglycerin‐induced migraine attack.Methods.—Thirty female migraineurs (without aura) and 11 controls participated in the study. Of these, 11 migraineurs and 5 controls practiced AT regularly for at least 6 months prior to and during the sublingual nitroglycerin test. Headache intensity and characteristics were recorded with a standardized method. During the nitroglycerin challenge, blood was collected for plasma cortisol determination and blood pressure and pulse rate were recorded.Results.—As a long‐term preventive treatment, AT significantly decreased the mean headache frequency and intensity (P= .001) compared to the pretreatment period in the migraineurs who regularly practiced AT (n = 11). During the nitroglycerin challenge, AT successfully attenuated the nitroglycerin‐induced acute decrease in blood pressure and pulse rate (P= .013; n = 16 AT subjects vs n = 25 non‐AT subjects). However, it was not effective in preventing immediate headache (P= .71), did not decrease the frequency of acute migraine attacks (P= .79), and could not alleviate acute migraine pain (P= .78; n = 16 AT subjects vs n = 25 non‐AT subjects). Plasma cortisol concentration significantly increased (P= .003) during the acute migraine attack (n = 22), and migraine intensity correlated with plasma cortisol elevations (P < .001; n = 41) and showed a tendency of negative correlation with morning plasma cortisol concentration (P= .08; n = 41). However, AT did not alter plasma cortisol responses (P= .99; n = 16 AT subjects vs n = 25 non‐AT subjects).Conclusion.—(1) The long‐term AT therapy proved to be a significantly effective preventive intervention in migraine sufferers. We hypothesized that this long‐term effect of AT is based on modulation of the pain anticipation system, which is strongly correlated with function of the anterior cingulate cortex. (2) We demonstrated that AT could not alter the nitroglycerin‐induced acute migraine attacks, which are directly related to the dysfunctional brainstem activation according to previous studies. (3) Our results suggested that there are multiple, complex relationships between cortisol responses and migraine pain, which are possibly mediated by the brain serotonergic system. (4) In addition, our results provide further evidence that nitroglycerin‐induced vasodilatation is not directly connected to either immediate headache or delayed migraine attack.