There is no consensus on the choice of drug treatment for refractory generalized convulsive status epilepticus in children. The objective of this meta-analysis of the published literature was to determine the effects of drug treatments on efficacy (seizure cessation) and mortality in children with this condition, controlling for potential confounding factors. One hundred eleven children, treated with diazepam, midazolam, thiopental, pentobarbital, or isoflurane, met strict inclusion criteria. Diazepam was significantly less efficacious than other treatments (P = .006) stratifying for etiology. Overall mortality was 20% in symptomatic cases and 4% in idiopathic cases (P = .038). Mortality was less frequent in midazolam-treated patients (P = .021) stratifying for etiology. Midazolam appears to be a good choice for initial treatment of refractory generalized convulsive status epilepticus in children, but the attribution of differences in efficacy and mortality solely to drug effect is not possible based on the published literature.