The mate choice behaviors of females can greatly affect patterns of reproductive success in males and influence the evolution of sexually selected male traits. Population-level estimates of display preferences may provide an accurate estimate of the strength and direction of selection by female choice if all females in the population show homogeneous preferences. However, population-level estimates may yield misleading estimates if there is within-population variation in mate preferences. While it is increasingly clear that the latter situation is common in nature, empirical data on the magnitude of variation in female preferences is required to improve our current understanding of its potential evolutionary consequences. We explored variation in female preference functions for three male call properties in a treefrog. We document substantial within-population variation not only in peak preferences but also in preference function shape (open, closed, flat), with at best 62% of females sharing a preference function shape with the respective population curve. Our findings suggest that population curves may accurately capture the direction of sexual selection, but depending on the properties of the constituting individual functions they may over- or underestimate the strength of selection. Particularly population estimates suggesting weak selection may in fact hide the presence of individual females with strong but opposing preferences. Moreover, due to the high within-population variation in both peak preferences and preference function shapes, the population functions drastically underestimate the predicted variation in male mating success in the population.