BACKGROUND: It has been estimated that about ten per cent of children between six and 20 years of age suffer from migraine. It is estimated that children with migraine lose one and a half weeks more schooling per year than their peers. Prophylactic drugs can be prescribed when children suffer from frequent or disabling headaches. OBJECTIVES: We aimed to describe and assess the evidence from controlled trials on the efficacy and tolerability of pharmacological agents taken on a regular basis to prevent the occurrence of migraine attacks and/or reduce the intensity of such attacks in children with migraine. SEARCH STRATEGY: The Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and EMBASE were searched from 1966 through 2002. Additional strategies for identifying trials included searching the reference lists of review articles and included studies and searching books related to headache. SELECTION CRITERIA: Prospective randomised controlled trials (RCTs) of self- or parent-administered drug treatments in children (under 18 years of age) who had received a diagnosis of migraine were included. DATA COLLECTION AND ANALYSIS: Two investigators extracted, assessed, and coded separately all data for each study, using a form that was designed specifically for the review. Any disagreement was resolved by discussion. Headache frequency standardised over 28 days was used as the primary outcome measure. Headache intensity, headache duration, amount of symptomatic treatment used, and headache indices were used as secondary outcome measures. Data were extracted from both parallel-group and crossover trials. Continuous and dichotomous data were used to calculate standardised mean differences (SMDs) and odds ratios (ORs), respectively. Numbers-needed-to-treat (NNTs) and numbers-needed-to-harm (NNHs) were also calculated. MAIN RESULTS: Thirty-eight studies were selected. Eighteen were excluded. Eleven preventive drugs were compared with placebo in a total of 15 studies. Drug-drug comparisons were made in just six studies. For only four drugs (L-5-hydroxytryptophan [L-5HTP], flunarizine, clonidine, and propranolol) were two or more studies selected. For only six drugs (trazodone, L-5HTP, propranolol, flunarizine, papaverine, and nimodipine) were data reported for effect on frequency. For no individual drug were comparable data reported in more than one study, thus meta-analysis was not possible. Two placebo-controlled studies showed a beneficial effect on the primary outcome measure, headache frequency. They were for the drugs propranolol and flunarizine. The propranolol study reported a dichotomous outcome (proportion of children responding), and it was possible to calculate a number-needed-to-treat to produce a two-thirds reduction in headache frequency (NNT = 1.5, 95%CI 1.15 to 2.1). The flunarizine study produced a SMD of 1.51 (95% confidence interval, -2.21 to -0.82), which was statistically significant in favour of flunarizine (p < 0.001). Nimodipine, timolol, papaverine, pizotifen, trazodone, L-5HTP, clonidine, metoclopramide, and domperidone showed no efficacy in reduction of frequency of attacks. The available studies on cyproheptadine, phenobarbitone, phenytoin, amitriptyline, carbamazepine, metoprolol, and piracetam were excluded for various reasons. REVIEWER'S CONCLUSIONS: Only one study each for propranolol and flunarizine were identified showing efficacy of these drugs as prophylactics of paediatric migraine. Nimodipine, timolol, papaverine, pizotifen, trazodone, L-5HTP, clonidine, metoclopramide, and domperidone showed no efficacy in reduction of frequency of attacks. Available studies on other commonly used drugs failed to meet our inclusion criteria. The quality of evidence available for the use of drug prophylaxis in paediatric migraine was poor. Studies were generally small, with no planning of sample size, so that for many drugs, despite the negative findings of this review, we do not have conclusive evidence of 'no effect'. There is a clear and urgent need for methodologically sound RCTs for the use of pings of this review, we do not have conclusive evidence of 'no effect'. There is a clear and urgent need for methodologically sound RCTs for the use of prophylactic drugs in paediatric migraine, starting with propranolol. These studies need to be adequately powered to investigate meaningful reductions in pain and suffering from a patient's perspective.