A combination of laboratory and field investigations of birds has shown that expression of behavior such as territorial aggression can occur throughout the year in many species and in different life history stages. Although it is well known that testosterone regulates territorial aggression in males during the breeding season, the correlation of plasma testosterone and aggression appears to be limited to periods of social instability when a male is challenged for his territory by another male, or when mate-guarding a sexually receptive female. How essentially identical aggression is modulated in non-breeding life history stages is not fully resolved, but despite low circulating levels of testosterone outside the breeding season, expression of territorial aggression does appear to be dependent upon aromatization of testosterone and an estrogen receptor-mediated mechanism. There is accumulating evidence that prolonged high levels of circulating testosterone may incur costs that may potentially reduce lifetime fitness. These include interference with paternal care, exposure to predators, increased risk of injury, loss of fat stores and possibly impaired immune system function and oncogenic effects. We propose six hypotheses to explain how these costs of high testosterone levels in blood may be avoided. These hypotheses are testable and may reveal many mechanisms resulting from selection to avoid the costs of testosterone. It should also be noted that the hypotheses are applicable to vertebrates in general, and may also be relevant for other hormones that have a highly specialized suite of actions in one life history stage (such as breeding), but also have a limited action in other life history stages when the full spectrum of effects would be inappropriate.