We followed-up 64 migraine patients after discontinuation of successful interval prophylaxis with flunarizine, propranolol or metoprolol, to investigate how long the therapeutic success would last, if further prophylaxis would be successful again, and what factors would influence the prognosis. We found that 16 out of 64 patients experienced a lasting reduction of migraine frequency, whereas 48 patients did improve initially, but later experienced a relapse. Further prophylaxis was effective in 29, poorly effective in 11, and ineffective in 8 of these patients; in 7 of the 8 non-responders prophylaxis was not changed. Negative prognostic factors were frequent attacks, a history of analgesic abuse and/or analgesic withdrawal therapy and ineffective previous prophylaxis. In conclusion, the therapeutic success decreases dramatically in the majority of patients several months after discontinuation of prophylaxis; further prophylaxis is more effective if the substance class is changed; increased analgesic intake is the most important prognostic factor. As a strategy for migraine prophylaxis we propose sequential changing of interval prophylaxis or--in patients with negative prognostic factors--long-term prophylaxis.