In this article. we present a thorough study on the autoxidation of cyclohexane, a model substrate for other (saturated) hydrocarbons. Despite the industrial impact of autoxidation reactions, a detailed mechanism is still missing. We present a combined experimental and computational study on the formation of both the major products (cyclohexylhydroperoxide, cyclohexanol and cyclohexanone), and the formation of ring-opened side-products. Up to now, these by-products, mainly adipic acid, were thought to originate from cyclohexanone. However, we found strong evidence that the subsequent propagation of ketone is much slower than assumed, and can only account for some 25% of ring-opened products. On the other hand, the hitherto completely overlooked propagation of the hydroperoxide, via fast alpha H-abstraction by chain-carrying peroxyl radicals, is identified as the major source of not only alcohol and ketone, but also by-products. In the case of N-hydroxyphthalimide (NHPI) catalysed oxidations, where mostly phthalimide N-oxyl (PINO center dot) radicals are propagating the chain, the situation is slightly different, as PINO center dot reacts more selectively with the alkane substrate than peroxyl radicals. This results in an increase in hydroperoxide selectivity. Lowering of the ROOH concentration by its, e.g. cobalt-catalyzed decomposition, leads to an enhanced catalytic efficiency, as a result of the shift in the ROO center dot + NHPI reversible arrow ROOH + PINO center dot equilibrium to the more efficient PINO center dot chain carrier. (c) 2006 Elsevier B.V. All rights reserved.