The purpose of this study was to determine whether a staged, algorithmic evaluation of infantile spasms could be developed to minimize patient discomfort, treatment delay, and overall costs. A retrospective chart review of patients diagnosed with infantile spasms at the authors' institution during a 10-year period was performed, with 29 patients identified; 28 charts were reviewed. By history and physical examination, 21 were classified as symptomatic and seven as cryptogenic. Of the 21 symptomatic patients, 13 had a known etiology at presentation; with further testing the specific etiology was determined in two more. Two in the cryptogenic group were reclassified on the basis of neuroimaging findings. Evaluations included neuroimaging (27, 15 abnormal), cerebrospinal fluid studies (nine, all normal), comprehensive metabolic studies (17, all normal), chromosomal analysis (11, two abnormal), and ophthalmologic evaluation (27, six abnormal). The average cost of the studies per patient was $5,076 at the authors' institution. Etiologic yield was increased by 20% with neuroimaging. The other investigations either confirmed a known etiology or were noncontributory. On the basis of these findings, the authors propose an algorithm for a more focused evaluation of infantile spasms. Using the algorithm, the authors suggest directly proceeding to therapy in patients with specific etiologies determined by history and examination. Further evaluation should start with neuroimaging. Subsequent evaluations should be on the basis of those results. The authors estimate a potential 60-90% reduction in total costs if this algorithm is applied.