This paper examines the case of natural hybridization between two angiosperm species (A and B) which are mechanically and ethologically isolated (or florally isolated). What is the effect of hybridization on the pollination system, and conversely, what is the effect of the pollinators on the outcome of the hybridization? The original floral isolation is based on an interspecific differentiation in floral characters, the floral mechanisms of the parental species being specialized for pollination by different types of pollinators with different body forms and behavioral traits. F1 hybrids of A x B have intermediate floral characters and serve as bridges for interspecific cross-pollination. The problem next shifts to the changes in floral characters and mode of pollination that are likely to occur in later generations in the hybrid population.The pollinators of species A and species B exert selective pressures on the hybrid population. If the normal pollinator of A is more abundant, active, and effective in the hybrid population than the pollinator of B, selection is expected to favor a reversion toward the floral characters and pollination system of A. The opposite condition, in which the pollinator of B is predominant, leads to the reciprocal result, reversion toward B. If the two types of pollinators are more or less the same in number of flower visits and pollination effectiveness, their combined selective pressure should produce latergeneration derivatives with intermediate floral characters suited for both pollinators. Three western North American plant groups containing florally isolated species that hybridize (Ipomopsis, Diplacus, and Aquilegia) are examined in relation to these predictions. The evidence in the three groups is generally in agreement with the hypothesis.Mechanical and ethological isolation often occur together in angiosperms and are referred to collectively as floral isolation. The floral isolation is frequently incomplete. A pair of florally isolated species may remain reproductively isolated in one area of sympatric contact but hybridize in another area. This pattern has been reported in a number of genera: Aquilegia (Ranunculaceae) (1-4), Epimedium (Berberidaceae) (5), Ipomopsis (Polemoniaceae) (4,(6)(7)(8), Penstemon (Scrophulariaceae) (9), Diplacus (Scrophulariaceae) (10), Rhinanthus (Scrophulariaceae) (11), Salvia (Labiatae) (12), and Platanthera (Orchidaceae) (13).We understand why the floral isolation breaks down locally. Floral isolation operates in conjunction with ecological isolation (4). Both modes of isolation are sensitive to manmade or natural disturbances of the original habitat differences between the related sympatric species. We have descriptive and analytical studies of the hybrids between florally isolated species in various plant groups. The initial hybridization is known to be followed by introgression in the species groups in six of the genera listed above: Aquilegia (1), Ipomopsis (6, 7), Diplacus (10, 14), Rhinanthus (11), Salvia (12), and Platanthera (13)....