Adenosine has long been considered an endogenous anti-epileptic compound. This concept was based on the widespread distribution of adenosine A 1 receptors (A 1 R), which are mostly located in excitatory synapses; here, A 1 R inhibit glutamate release, decrease glutamatergic responsiveness and hyperpolarise neurons. However, the combined observation that synaptic A 1 R undergo desensitisation in chronic noxious situations whereas the activation of A 1 R still prevents seizure activity suggests that the A 1 R anti-epileptic action may involve non-synaptic mechanisms. Two alternative mechanisms can be considered to explain the ability of A 1 R to control seizure activity and resulting neurodegeneration: 1) the possible role of A 1 R-mediated control of metabolism; 2) the A 1 R-mediated preconditioning involving a coordinated control of neuron-glia communication. However, purinergic modulation of seizure activity is likely to involve other systems apart from A 1 R. Thus, the blockade of adenosine A 2A receptors (A 2A R), which density increases in animal models of epilepsy, can attenuate seizure activity and prevent seizure-induced neurodegeneration. Furthermore, ATP, which is the main source of the endogenous adenosine activating A 2A R, also act as a general danger signal and may also directly control seizure activity through P 2 receptors (P 2 R). Therefore, the purinergic control of epilepsy may actually involve different parallel signalling arms, some beneficial and others deleterious, probably acting at different sites (in epileptic foci and in their neighbourhood) and at different times. It is likely that combined targeting of different purinergic receptors may be the most efficacious way to control seizure activity, its spreading and the resulting neurodegeneration.