A cobalt chloride solution was injected into the unilateral sensorimotor cortex of rats to induce epileptic activity. The cyclic AMP contents of slices incubated with or without adenosine and 2-chloroadenosine were determined in four cortical areas after electroencephalography and behavioral examination in cobalt-injected rats. Electrographic spike activity appeared immediately after injection of cobalt. In the majority of cobalt-injected rats, the spike activity was dominant in the primary epileptic region of the cortex. The spike frequency reached a maximum level two to three weeks after the injection and declined thereafter. The electrographic activity was followed by abnormal behavior. Adenosine and 2-chloroadenosine elevated the cyclic AMP levels in the cortical slices 6-to 10-fold and 10-to 16-fold, respectively. The elicitation of cyclic AMP accumulation was strongly inhibited by the adenosine antagonist 8-phenyltheophylline. The cyclic AMP accumulation elicited by adenosine or 2-chloroadenosine was increased in the primary cortical area of cobalt-induced epilepsy, but in the other cortical areas there was no deviation in cyclic AMP accumulation. The increase in cyclic AMP accumulation was observed regardless of the presence or absence of the adenosine uptake inhibitor dipyridamole, phosphodiesterase inhibitor Ro 20-1724, and adenosine deaminase. The increased accumulation of cyclic AMP in the primary epileptic cortex was detected as early as 8 days after the injection. The cyclic AMP accumulation slightly increased thereafter. It reached a plateau 17 to 19 days after the injection and then turned to the control levels, in harmony with the electrographic and behavioral profiles. These findings suggest that alterations in adenosine-sensitive generation of cyclic AMP in the primary epileptic region of the cortex are part of the neurochemical process of cobalt-induced epilepsy.