Vertebrates show two major classes of sexually dimorphic traits: weaponry and ornaments. However, Darwin could not explain why their expression varies so much across lineages. We argue that coercion-avoidance can explain both the existence and taxonomic distribution of ornaments. Females maximize their fitness when they can freely choose their mates, but males are expected to use sexually dimorphic weaponry not only to displace other males, but also to overcome female preferences and thus acquire matings by force whenever they can. Females should therefore avoid coercive males and avoid using weaponry as a criterion for male quality wherever possible, and rely on male viability indicators that cannot be used to coerce females (i.e. ornaments). Ornaments predominate in birds and weaponry in mammals because female choice is less costly in birds, due to higher intrinsic female behavioural freedom and lower male monopolization potential. We also predict that specialized coercive organs occur where females have low behavioural freedom but males benefit little from weaponry in male–male contests. A review of the empirical evidence supports the basic predictions of this coercion-avoidance hypothesis. We also present a simple mathematical model that confirms the logic of this hypothesis.