The object of this work was to subject established empirical medical treatment regimens for infantile spasms to evidence-based medicine analysis in order to determine the current best practice for the treatment of infantile spasms in children. Clinical studies of infantile spasms reported during the presteroid era were reviewed critically to define the natural history of the disorder. Treatment trials of infantile spasms conducted since 1958 were rigorously assessed using MEDLINE and hand searches of the English language literature. Inclusion criteria were the documented presence of infantile spasms and hypsarrhythmia. Outcome measures included complete cessation of spasms, resolution of hypsarrhythmia, relapse rate, developmental outcome, the presence or absence of epilepsy, and/or an epileptiform electroencephalogram. Evidence was defined as class I, II, or III, and practice parameter recommendations were made using the framework devised by the American Academy of Neurology. Class I and III evidence support a standard of practice recommendation for the use of vigabatrin in the treatment of infantile spasms in children with tuberous sclerosis. Class I and III evidence support a guidelines recommendation for the use of either ACTH or vigabatrin in infantile spasms in nontuberous sclerosis patients. There is no strong evidence that successful treatment of infantile spasms improves the long-term prognosis for cognitive outcome or decreases the incidence of later epilepsy. A practice option recommendation for the use of oral corticosteroids in the treatment of infantile spasms is supported by limited and inconclusive class I and III data. Based on the evidence, no recommendation can be made for the use of pyridoxine, benzodiazepines, or the newer antiepileptic drugs in the treatment of infantile spasms. ACTH and vigabatrin are the most effective agents in the treatment of infantile spasms, but concerns remain about the risk/benefit profiles of these drugs.