BACKGROUND: Anticonvulsant drugs seem to be useful in clinical practice for the prophylaxis of migraine. This might be explained by a variety of actions of these drugs in the central nervous system that are probably relevant to the pathophysiology of migraine. OBJECTIVES: To describe and assess the evidence from controlled trials on the efficacy and tolerability of anticonvulsants for preventing migraine attacks in adult patients with migraine. SEARCH STRATEGY: We searched MEDLINE (from 1966 on) and the Cochrane Central Register of Controlled Trials (CENTRAL). Date of most recent search: April 2003. Additional information was gained from hand-searching specialist headache journals; correspondence with pharmaceutical companies, authors of reports, and experts in the field; and a wide variety of review articles and book chapters. SELECTION CRITERIA: Studies were required to be prospective, controlled trials of self-administered drug treatments taken regularly to prevent the occurrence of migraine attacks and/or to reduce the intensity of those attacks. DATA COLLECTION AND ANALYSIS: Studies were selected and data extracted by two independent reviewers. For migraine frequency data, standardized mean differences (SMDs) were calculated for individual studies and pooled across studies. For dichotomous data on significant reduction in migraine frequency, odds ratios (ORs) and numbers-needed-to-treat (NNTs) were similarly calculated. Adverse events were analyzed by calculating numbers-needed-to-harm (NNHs) for studies using similar agents. MAIN RESULTS: Fifteen papers were included in the review. Of these, 14 reported trials comparing anticonvulsants with placebo, as follows: four trials of divalproex sodium, three trials of topiramate, two trials of sodium valproate, two trials of gabapentin, and one trial each of carbamazepine, clonazepam, and lamotrigine. One paper reported a trial of sodium valproate versus an active comparator, flunarizine, and one trial of divalproex sodium versus placebo included a comparison against propranolol, also an active comparator. Data from 2024 patients were considered. Analysis of data from eight trials (n = 841) demonstrates that anticonvulsants, considered as a class, reduce migraine frequency by about 1.4 attacks per 28 days as compared to placebo (SMD -0.60; 95% confidence interval [CI] -0.93 to -0.26). Data from 10 trials (n = 1341) show that anticonvulsants, considered as a class, also more than double the number of patients for whom migraine frequency is reduced by 50% or more, relative to placebo (OR 3.90; 95% CI 2.61 to 5.82; NNT 3.8; 95% CI 3.2 to 4.6). For seven trials of sodium valproate and divalproex sodium, NNHs for five clinically important adverse events ranged from 6.6 to 16.3. For the three trials of topiramate, NNHs for eight adverse events (100-mg dose) ranged from 2.4 to 32.9. REVIEWERS' CONCLUSIONS: Anticonvulsants appear to be both effective in reducing migraine frequency and reasonably well tolerated. There is noticeable variation among individual agents, but there are insufficient data to know whether this is due to chance or variation in true efficacy. Neither clonazepam nor lamotrigine was superior to placebo (one trial each). Relatively few robust trials are available for agents other than sodium valproate/divalproex sodium. Two recently published and large trials of topiramate demonstrated reasonable efficacy, and one further trial of this agent is anticipated in the near future.