Neuronal cultures from rat cerebellum consisting of approximately 90% glutamatergic granule neurons, 5-7% GABAergic inhibitory interneurons, and 3-5% glial cells, were treated for four days with 50 microM kainic acid (KA) to determine the cellular origin of released endogenous neuroactive substances. KA, known to be selectively toxic to GABAergic neurons, caused an estimated 80% decrease in glutamic acid decarboxylase (GAD) immunofluorescence. Furthermore, K(+)-stimulated release of GABA decreased to 20% of control values, and did not return to control levels in cultures "recovered" two days in KA-free media, suggesting the loss of inhibitory interneurons. Similarly, adenosine and taurine showed decreased K(+)-stimulated release, which was unrecoverable when KA was removed from the medium. K(+)-stimulated release of glutamate and aspartate also decreased by 50% and 70%, respectively, after chronic KA treatment. In contrast, however, this release returned to control levels in recovered cultures. All decreases in K(+)-stimulated release were prevented by concurrent treatment with KA and the KA antagonist 6-cyano-6-nitroquinoxaline-2,3-dione (CNQX), indicating that a receptor-mediated mechanism was involved. We conclude that, in these cultures, most of the K(+)-stimulated release of adenosine and taurine originates from the GABAergic interneurons, the basket and stellate cells, which are selectively killed by the KA treatment. The data also strongly suggest that glutamate and aspartate, the levels of which recover after KA treatment, originate mainly from the granule neurons.