In an incidence cohort, remission and relapse rates and determinants were studied in 178 patients followed long-term. A comparative study of predictive factors was performed in 40 patients with histories of antiepileptic (AED)-drug-refractory epileptic seizures in the last 10 years of follow-up and compared with the other 138 cohort subjects. The two groups were cross-tabulated with 353 variables of family history, obstetric, developmental and seizure histories, and current medical and social status. Multivariate analyses were applied for control of confounding. Defined or probable remote symptomatic etiology of seizures, abnormal neurologic development/status, high initial seizure frequency, occurrence of status epilepticus, and poor short-term effects of AED therapy were significantly associated with long-term AED refractoriness. On logistic regression analyses, poor short-term outcome of AED therapy [odds ratio (OR) 3.6; 95% confidence interval (CI) 1.2-10.4], occurrence of status epilepticus (OR 11.4; 95% CI 3.2-41.0), high initial seizure frequency (OR 4.6; 95% CI 1.1-19.3), and remote symptomatic seizure etiology (OR 2.9; 95% CI 1.1-8.2) remained the only independent predictors of seizure intractability. These factors enable early assessment of need for epilepsy surgery.