Most pollinators visit flowers in the search of nectar rewards. However, as the floral nectar can often not be directly detected by pollinators, many flower visitors use secondary metabolites such as odor- or taste-proxies to anticipate nectar quantity and quality. Plants might exploit these sensory inferences of the pollinator to increase their pollination rates without increasing their caloric investment into their floral rewards. Here we investigated the effects of natural variation in certain primary and secondary floral metabolites in three populations of the wild tobacco, Nicotiana attenuata, on the pollination behavior of the hawkmoth Manduca sexta. Although offering the same caloric value per flower, the plants of these populations differ in the compositions and concentrations of sugars within the nectar. Moreover, the flowers of these plants emitted highly contrasting levels of attractive floral volatiles (benzyl acetone), but did not differ in the amounts of defensive nectar metabolites (nicotine). In wind tunnel assays with M. sexta moths, plants from those populations that released the largest amount of benzyl acetone as well as those that had a higher ratio of nectar sucrose were more frequently visited and re-visited by the hawkmoth. High emissions of benzyl acetone additionally correlated with a higher time investment of the moths into individual flowers on each visit, leading to the largest foraging success of the moths on those flowers that were most strongly scented. We propose that it is the variation of flower metabolites and their detection by the pollinator rather than the actual caloric value of the nectar, which determines pollinator visitations to a certain flower population. Hence, plants could potentially create a specialist pollinator community by altering their floral signals, either by producing volatiles that pollinators prefer or by providing nectar sugars that pollinators are most sensitive to, while at the same time keeping the caloric value of their nectar rewards at a constant level. These findings highlight the importance of variations among floral signals and their detection by specific pollinators for plant diversification as well as for the evolution of specific plant-pollinator interactions.