Refractory status epilepticus (SE) has a mortality rate of up to 35%. Current treatment protocols for the treatment of SE begin with benzodiazepines and then proceed to conventional anticonvulsants. If seizures continue, SE is considered refractory (RSE) and treatment with anesthetic agents in undertaken. Twenty-four h to 48 h after initiation of anesthesia with midazolam, pentobarbital or thiopental, or propofol, an attempt is made to wean the anesthetic. If this fails and seizures recur, SE is considered highly refractory (HRSE) and repeated attempts are undertaken. No randomized trial data are available to guide the choice of anesthetic agent in either RSE or HRSE status. Medication resistance in established SE is thought to result, in part, from internalization of synaptic γ-aminobutyric acid (GABA) receptors, making them unavailable for modulation. Neurosteroids act on both synaptic and extrasynaptic GABAA receptors, which are not internalized, and are therefore hypothesized to have a role in the treatment of RSE. SGE-102 is a neurosteroid metabolite of progesterone with demonstrated anticonvulsant properties in animal seizure models. A randomized double-blind placebo-controlled adjunctive trial of SGE will include subjects randomized at the time that initial treatment with anesthesia is initiated. Subjects will receive midazolam and either SGE-102 or placebo. Midazolam will be tapered and discontinued between hours 24 and 48. SGE-102 or placebo will be continued through hour 120. The primary end point will be the difference in proportion of subjects from each arm who remain seizure free through hour 120. Secondary end points will include the proportion of subjects who are seizure free at hour 168, 2 days after discontinuation of the experimental agent. The study will be powered to have a 90% chance of detecting a clinically meaningful reduction in seizure recurrence at 120 h. Comprehensive safety and pharmacokinetic data will also be obtained during the course of the trial.