For many migraine patients, triptan therapy provides complete pain relief in some attacks but not in others. Here, we tested whether the success of triptan therapy is hindered in the presence of cutaneous allodynia (pain resulting from a nonnoxious stimulus to normal skin), a phenomenon we previously described develop gradually during the course of the migraine attack in more than 70% of patients. We studied migraine patients repeatedly on three visits to the clinic: in the absence of migraine (baseline), within the first hour of one attack, or at 4 hours from onset of another attack. Presence or absence of allodynia was determined based on differences between migraine and baseline pain thresholds to mechanical and thermal stimulation of periorbital skin. In 31 patients, we studied 34 migraine attacks that were associated with allodynia at the time of triptan treatment and 27 attacks that were not. Within 2 hours of triptan treatment, patients were rendered pain-free in 5 of 34 (15%) of allodynic attacks versus 25 of 27 (93%) of nonallodynic attacks. Treating migraine attacks 1 hour (early) or 4 hours (late) after the onset of pain was equally ineffective in inducing a pain-free state in the presence of allodynia, and equally effective in the absence of allodynia. For patients susceptible to allodynia during the attack, triptan therapy was by far more likely to provide complete pain relief if administered before rather than after the establishment of cutaneous allodynia. Patients who never developed allodynia were highly likely to be rendered pain-free by triptan therapy anytime after the onset of pain. We conclude that the probability of consistent pain-free outcome increases drastically if triptan therapy is vigilantly timed to precede any signs of cutaneous allodynia. Ann Neurol 2004;55:000000