Patients aged 19 to 65 years whose headache was severe enough to require parenteral therapy and who met International Headache Society migraine criteria were eligible for this randomized, double-blind trial. The study was conducted in the ED of 2 community hospitals, 1 of which was a tertiary referral centre. Exclusion criteria included pregnancy, focal findings, fever, meningismus, allergy to the study drug, active peptic ulcer disease and diabetes mellitus. Demographic and clinical data, including headache severity, were recorded. After abortive therapy (antiemetics, intravenous nonsteroidal agents, dihydroergotamine or opioids), blinded nurses administered dexamethasone (24 mg intravenously) or placebo. Patients recorded headache severity on a Visual Analogue Scale (VAS) at time T = 0, T = 30 minutes and T = 60 minutes and at discharge. They were contacted 48 to 72 hours later and asked whether they had suffered a recurrence of their headache, categorized as class A (severe, provoking another physician visit), class B (severe, interfering with daily activity but not provoking a physician visit), class C (mild, requiring self-medication but not limiting activity) or class D (mild, requiring no treatment). Results: Two of 100 patients were lost to follow-up, leaving 98 in the study sample. Placebo recipients were more likely to be female; other baseline characteristics were similar between groups. Median VAS pain score was 83 mm on ED arrival, 35 mm after initial treatment and 12 mm on discharge. At follow-up, 65 of 98 patients had suffered headache recurrence. In the placebo versus dexamethasone groups, respectively, the results were 11 versus 0 in class A, 11 versus 9 in class B, 7 versus 11 in class C and 4 versus 12 in class D. Regarding the primary outcome, 9 of 49 dexamethasone patients (18%) and 22 of 49 placebo patients (45%) had severe (classes A and B) recurrent headache (odds ratio 0.28; 95% CI, 0.11 to 0.69; p = 0 .005).