Selection mediated by one biotic agent will often be modified by the presence of other biotic interactions, and the importance of such indirect effects might change over time. We conducted an 11-year field experiment to test the prediction that large grazers affect selection on floral display of the dimorphic herb Primula farinosa not only directly through differential grazing damage, but also indirectly by affecting vegetation height and thereby selection mediated by pollinators and seed predators. Exclusion of large grazers increased vegetation height and the strength of pollinator-mediated selection for tall inflorescences and seed predator-mediated selection for short inflorescences. The direct effect of grazers on selection resulting from differential grazing damage to the two scape morphs showed no temporal trend. By contrast, the increase in vegetation height in exclosures over time was associated with an increase in selection mediated by pollinators and seed predators. In the early years of the experiment, the indirect effects of grazers on selection mediated by pollinators and seed predators were weak, whereas at the end of the experiment, the indirect effects were of similar magnitude as the direct effect due to differential grazing damage. The results demonstrate that the indirect effects of a selective agent can be as strong as its direct effects, and that the relative importance of direct vs. indirect effects on selection can change over time. A full understanding of the ecological processes governing variation in selection thus requires that both direct and indirect effects of biotic interactions are assessed.