To understand seizures and the enduring predisposition of the brain to generate seizures, it is crucial to elucidate the dynamical pathways through which the brain reaches the seizure state. Results from computational modelling studies have predicted the existence of specific pathways to seizure. In our study, we have explored whether the transition to seizure follows principles of a universal dynamical process in nature – critical slowing. Seizures were induced in rat hippocampal brain slices by perfusion of the slices with high potassium artificial cerebrospinal fluid. Chronic epilepsy in rats was induced by the injection of a minute dose of tetanus toxin to the right dorsal hippocampus. In all preparations, we recorded spontaneous or electrically induced brain activity. The local field potentials were analyzed and temporal profiles of early warning signals of critical slowing determined. Long‑term human intracranial recordings were obtained from patients with an implanted seizure monitoring device. Combining the experimental approaches in in vitro and in vivo models of seizures with the analysis of long-term recordings in patients, we were able to demonstrate that the transition to seizure is associated with changes in neuronal and network activity which displays features of critical slowing – a dynamical phenomenon which reflects progressive loss of brain stability. With approaching seizure, the brain became more sensitive to internal and external perturbations, and it was also characterized by delayed recovery from the perturbations. Our results suggest that transition to seizure may be a slowly changing process characterized by discrete changes in brain dynamics. Also, the loss of resilience has the potential to be better controlled than the random occurrence of stochastic perturbations initiating a seizure.