We applied nonlinear analysis to the results of electroencephalography (EEG) in a pilocarpine-induced status epilepticus (SE) model to characterize nonlinear dynamics according to SE phase. Nine male Sprague-Dawley rats weighing 150-250 g were used. EEG was classified into four phases in addition to baseline EEG (phase 0) as follows: phase 1, discrete seizures; phase 2, continuous ictal discharges; phase 3, early periodic epileptiform discharges (PEDs); and phase 4, late PEDs. High-dose diazepam was administered at phase 4 to terminate SE. Diazepam controlled SE in five rats (group 1), while it failed to stop SE in the rest (group 2). The presence of nonlinearity was determined by time reversal asymmetry statistics using a surrogate data set. The correlation dimension (D(2)) was calculated to characterize the dimensional complexity of each phase of SE. EEG of later phases of SE showed strong nonlinearity, whereas no or only weak nonlinearity was noted at phases 0 and 1. D(2) showed the highest value at phase 0 and decreased progressively. Considering therapeutic responsiveness, D(2) showed significant differences between the two groups at phases 2 and 4. These results suggest that nonlinear dynamic changes in the later SE phases reflect underlying pathophysiological changes that contribute to determining therapeutic responsiveness in the pilocarpine-induced SE model.