AIM To identify predictors of seizure control in newly presenting children with epilepsy in countries with limited resources.METHOD Three hundred and ninety children (273 males, 117 females) aged 2 months to 15 years with newly diagnosed epilepsy were enrolled prospectively at first visit to the multidisciplinary clinic at the children's hospital in Dhaka, Bangladesh. Data about seizures, motor disability, psychomotor development, and electroencephalography were obtained. Regular monitoring of antiepileptic drug treatment was continued at least for one year. Associations between seizure control and potential predictors were determined by multivariate analysis.RESULTS Three hundred and ninety children were enrolled in 6 months, of whom over 60% were from low-income families, 60% had onset at under 1 year, 74% had more than one seizure per week, 69% a single-seizure type, and 38% a history of delayed onset of breathing at birth. Cognitive deficits (IQ<70; 58%) and ⁄ or motor (significant limitation of daily living activities; 47%) deficits were common. After 1 year of regular treatment, seizure control was good (seizure freedom) in 53%, and poor (at least one seizure in the last 3mo of follow-up) in 47%. The predictors of poor seizure control were an IQ<70, associated motor disability, multiple seizure types, and a history of cognitive regression (1.9 times more likely to have poor seizure control).INTERPRETATION Seizure control can be predicted using three clinical factors (motor disability, cognitive impairment, and multiple seizure types) at the first clinic visit. Such predictors assist the development of referral plans and management guidelines for childhood epilepsies in resourcepoor countries.The management of childhood epilepsy in countries with limited resources has several specific problems. These include lack of medical resources, misunderstanding of the nature of epilepsy, stigmatization, large treatment gaps (i.e. a high percentage of the target population not receiving medication but often a high level of use of traditional healers), and a lack of understanding of the priorities of the users by those seeking to provide treatment. All of these issues arise in the management of epilepsy in children in Bangladesh. In consultation with parents, we formed the view that the ability to predict seizure control with future treatment can be useful. First, it provides an audit so that if the control is widely discrepant from that predicted, a reason may be sought (e.g. seizures were not epileptic, antiepileptic drugs were not taken). Second, it gives the family and patient an expectation of what antiepileptic drugs can achieve and whether they are essential to management. Third, it allows an approximate cost-benefit analysis to be performed. These may be useful in any setting, but in a country with limited resources (e.g. Bangladesh) this may allow doctors and carers to deploy their resources more effectively. The second broad issue raised was the high rate of additional problems of development and behaviour th...