I review the hypotheses that have been proposed to explain the adaptive value of autumn leaf colours. The available adaptive hypotheses can be reduced to the following. Photoprotection: pigments protect against photoinhibition or photooxidation allowing a more efficient recovery of nutrients. Drought resistance: pigments decrease osmotic potential allowing leaves to tolerate water stress. Leaf warming: pigments convert light into heat and warm leaves. Fruit flag: colour attracts animals that help disperse seeds. Coevolution: colour signals that the tree is not a suitable host for insects. Camouflage: colour makes leaves less detectable to herbivores. Anticamouflage: colour enhances conspicuousness of parasites dwelling on leaves to predators or parasitoids. Unpalatability: pigments act as direct anti-feedants against herbivores. Reduced nutrient loss: yellow leaves have less to lose against herbivory. Tritrophic mutualism: colour attracts aphids which attract ants that defend the trees from other insects. For each hypothesis I mention the original references, I define assumptions and predictions, and I discuss briefly conceptual problems and available evidence.