Current therapies for epilepsy are largely symptomatic and do not affect the underlying mechanisms of disease progression, i.e. epileptogenesis. Given the large percentage of pharmacoresistant chronic epilepsies, novel approaches are needed to understand and modify the underlying pathogenetic mechanisms. Although different types of brain injury (e.g. status epilepticus, traumatic brain injury, stroke) can trigger epileptogenesis, astrogliosis appears to be a homotypic response and hallmark of epilepsy. Indeed, recent findings indicate that epilepsy might be a disease of astrocyte dysfunction. This review focuses on the inhibitory neuromodulator and endogenous anticonvulsant adenosine, which is largely regulated by astrocytes and its key metabolic enzyme adenosine kinase (ADK). Recent findings support the "ADK hypothesis of epileptogenesis": (i) Mouse models of epileptogenesis suggest a sequence of events leading from initial downregulation of ADK and elevation of ambient adenosine as an acute protective response, to changes in astrocytic adenosine receptor expression, to astrocyte proliferation and hypertrophy (i.e. astrogliosis), to consequential overexpression of ADK, reduced adenosine and - finally - to spontaneous focal seizure activity restricted to regions of astrogliotic overexpression of ADK. (ii) Transgenic mice overexpressing ADK display increased sensitivity to brain injury and seizures. (iii) Inhibition of ADK prevents seizures in a mouse model of pharmacoresistant epilepsy. (iv) Intrahippocampal implants of stem cells engineered to lack ADK prevent epileptogenesis. Thus, ADK emerges both as a diagnostic marker to predict, as well as a prime therapeutic target to prevent, epileptogenesis.