Medical records of 60 children (53% males) with cryptogenic epilepsy, seen over a 12-year period and treated at Montreal Children’s Hospital, McGill University, Canada, were reviewed retrospectively to identify clinical and predictive features of outcome. Mean age of initial seizure was 7.7 years (range 2.3-16.4 years). Febrile seizures had preceded the onset of epilepsy in 11 (18.3%), of which 7 were simple in type and 4 complex. Seizure types at presentation were generalized in 22 (35.7%), only 1 a typical absence; partial with secondary generalization in 20 (33.3%), complex partial in 17 (28.3%), and simple partial (1). Physical examination was normal in 46 (76.7%), 9 (15%) were developmentally delayed at onset, 38 (63.3%) attended regular classes, 11 (18.3%) had ADHD and 12 (20%) had a learning disability at onset. After initiating antiepileptic medication, mean follow-up was 53 months (range 24-128 months). Seizures were completely controlled for at least 24 months in 29 (48.3%), patients were seizure free for >12 months in 44 (73.3%), seizures recurred within 12 months in 16 (26.7%), and were intractable in 4 (6.7%). Outcome was poor in 8 (13.3%) and very poor in 4 (6.7%). Risk factors for a poor outcome were a seizure recurrence within 6-12 months after initiating therapy (p=0.006), and developmental delay at onset (p=0.023). 
COMMENT. Cryptogenic epilepsy refers to seizures without genetic or known cause but presumed to be symptomatic, possibly a subtle cerebral dysgenesis undetected by MRI. The term idiopathic is reserved for seizures solely genetic in origin. Cryptogenic epilepsy in childhood generally has a favorable outcome, approximately 50% being seizure-free for more than 2 years after starting AEDs. Risk factors for a poor outcome include seizure recurrence within 6-12 months of treatment initiation and developmental delay at seizure onset. As expected, seizure intractability and poor outcome are correlated with the need for addition of a second AED. Except for developmental delay, initial clinical features are not helpful in predicting outcome. Despite the frequency of ADHD and learning disabilities, more than half the patients do not require special classes. Normal neurologic examination and the absence of comorbid disorders at onset of seizures do not always correlate with a favorable outcome, 55% requiring more than one AED. Seizure recurrences and developmental delay indicate a need for aggressive therapy.