BACKGROUND: In endemic areas, falciparum malaria remains the leading cause of seizures in children presenting to emergency departments. In addition, seizures in malaria have been shown to increase morbidity and mortality in these patients. The management of seizures in malaria is sometimes complicated by the refractory nature of these seizures to readily available anti-convulsants. The objective of this study was to determine predictors of anti-convulsant treatment failure and seizure recurrence after initial control among children with malaria. METHODS: In a previous study, the efficacy and safety of buccal midazolam was compared to that of rectal diazepam in the treatment of prolonged seizures in children aged three months to 12 years in Kampala, Uganda. For this study, predictive models were used to determine risk factors for anti-convulsant treatment failure and seizure recurrence among the 221 of these children with malaria. RESULTS: Using predictive models, focal seizures (OR 3.21; 95% CI 1.42-7.25, p = 0.005), cerebral malaria (OR 2.43; 95% CI 1.20-4.91, p = 0.01) and a blood sugar >or=200 mg/dl at presentation (OR 2.84; 95% CI 1.11-7.20, p = 0.02) were independent predictors of treatment failure (seizure persistence beyond 10 minutes or recurrence within one hour of treatment). Predictors of seizure recurrence included: 1) cerebral malaria (HR 3.32; 95% CI 1.94-5.66, p < 0.001), 2) presenting with multiple seizures (HR 2.45; 95% CI 1.42-4.23, p = 0.001), 3) focal seizures (HR 2.86; 95% CI 1.49-5.49, p = 0.002), 4) recent use of diazepam (HR 2.43; 95% CI 1.19-4.95, p = 0.01) and 5) initial control of the seizure with diazepam (HR 1.96; 95% CI 1.16-3.33, p = 0.01). CONCLUSION: Specific predictors, including cerebral malaria, can identify patients with malaria at risk of anti-convulsant treatment failure and seizure recurrence.