We report the long-term outcome of 64 infants with infantile spasms, followed prospectively, using controlled treatment schedules and objective techniques (24-hour EEG and video monitoring) to determine response. Average age at follow-up was 50 months. Of the 64 infants, three (5%) died; of the others, 41 (67%) had developmental retardation of 50% or more or an IQ of 50 or less. Eight patients (13%) composed our cryptogenic study group and were so classified on the basis of normal CT scan, normal development prior to onset of infantile spasms, and undetermined cause. These patients had the better outcome; 38% had normal development or were only mildly retarded. Both the responders and nonresponders in our symptomatic group had a poor outcome; only 5% had normal development or mild impairment. Outcome was not significantly influenced by short versus long treatment lag or by response to therapy. Other types of seizures occurred in 34 patients (53%). In summary, the overall prognosis for long-term outcome in these 64 patients with infantile spasms was poor.