We have found previously that astrocytes can provide cysteine to neurons. However, cysteine has been reported to be neurotoxic although it plays a pivotal role in regulating intracellular levels of glutathione, the major cellular antioxidant. Here, we show that cysteine toxicity is a result of hydroxyl radicals generated during cysteine autoxidation. Transition metal ions are candidates to catalyze this process. Copper substantially accelerates the autoxidation rate of cysteine even at submicromolar levels, whereas iron and other transition metal ions, including manganese, chromium, and zinc, are less efficient. The autoxidation rate of cysteine in rat CSF is equal to that observed in the presence of similar to0.2 muM copper. In tissue culture tests, we found that cysteine toxicity depends highly on its autoxidation rate and on the total amount of cysteine being oxidized, suggesting that the toxicity can be attributed to the free radicals produced from cysteine autoxidation, but not to cysteine itself. We have also explored the in vivo mechanisms that protect against cysteine toxicity. Catalase and pyruvate were each found to inhibit the production of hydroxyl radicals generated by cysteine autoxidation. In tissue culture, they both protected primary neurons against cysteine toxicity catalyzed by copper. This protection is attributed to their ability to react with hydrogen peroxide, preventing the formation of hydroxyl radicals. Pyruvate, but not catalase or glutathione peroxidase, was detected in astrocyte-conditioned medium and CSF. Our data therefore suggest that astrocytes can prevent cysteine toxicity by releasing pyruvate.