It is suggested that the functional significance of each di/trimorphic structure in heterostyled flowers should be considered separately. The origin of the morphological and physiological components of heterostyly is discussed; it is suggested that the physiological incompatibility arose first and created conditions favouring selection of morphological differentiation. In this connection, the cases of Amsinckia and Narcissus tazetta are given special consideration. The restricted occurrence of tristyly is considered together with the distinctive syndrome of exserted sexual organs and sternotribic pollination in heterostyled plants. The need is emphasized to bear in mind constantly the functional aspects of heterostyly.