Benzodiazepines are the most potent drugs used in the management of status epilepticus (SE). A number of presynaptic, postsynaptic, and nonsynaptic actions of benzodiazepines have been described. However, only the benzodiazepines' enhancement of gamma-aminobutyric acid (GABA)ergic inhibition and their reduction of repetitive firing occur at concentrations of unbound drug comparable to those that block absence seizures or stop clinical SE in patients. Thus, it is likely that these actions contribute to antiepileptic and anti-SE efficacy of the benzodiazepines. A predictable sequence of progressive electroencephalographic (EEG) changes during the course of generalized convulsive SE, both in humans and in experimental models, has been recently described. The homology of the sequence of EEG patterns in patients and in experimental models supports the concept that animal models should be useful in evaluating the treatment of clinical SE, and benzodiazepines are effective in stopping SE in a number of animal models. Late SE in animals, however, as in humans, is less responsive to treatment than is early SE. Forty-seven clinical studies in which clonazepam, diazepam, or lorazepam was used in the treatment of SE have been reported. Overall, lasting control of SE was achieved in 79% of the 1,346 patients in these noncontrolled studies. However, no data yet exist to differentiate the efficacy of 1 of the benzodiazepines from that of the others. Therefore, the choice of benzodiazepine is best determined by availability and by pharmacokinetic differences. Because of a much smaller volume of distribution of unbound drug, lorazepam appears to have a significantly longer effective duration of action against SE than does diazepam, which is rapidly redistributed to lipid stores in the body after intravenous administration. For this reason, we now use lorazepam in the initial treatment of patients with generalized convulsive SE.