Biologists have long assumed that pollinator behavior is an important force in angiosperm speciation, yet there is surprisingly little direct evidence that floral preferences in pollinators can drive floral divergence and the evolution of reproductive (ethological) isolation between incipient plant species. In this study, we expose computer‐generated plant populations with a wide variation in flower color to selection by live and virtual hummingbirds and bumblebees and track evolutionary changes in flower color over multiple generations. Flower color, which was derived from the known genetic architecture and phenotypic variance of naturally occurring plant species pollinated by both groups, evolved in simulations through a genetic algorithm in which pollinator preference determined changes in flower color between generations. The observed preferences of live hummingbirds and bumblebees were strong enough to cause adaptive divergence in flower color between plant populations but did not lead to ethological isolation. However, stronger preferences assigned to virtual pollinators in sympatric and allopatric scenarios rapidly produced ethological isolation. Pollinators can thus drive ecological speciation in flowering plants, but more rigorous and comprehensive behavioral studies are required to specify conditions that produce sufficient preference levels in pollinators.