In leks, females are often attracted to males bearing costly secondary sexual traits that are thought to honestly signal male genetic quality. The Good Genes Hypothesis predicts 1) consistent male attractiveness across a variety of environments, 2) directional selection on male attractiveness, and 3) a positive correlation between paternal attractiveness and offspring fitness. We tested these predictions in a natural population of the Italian tree frog, Hyla intermedia. We recorded the calling activity of all males at a small lek for 2 breeding seasons and we carried out an incomplete factorial breeding experiment to test for an association between paternal attractiveness and offspring fitness. Because we found no evidence for temperature-dependent genotype-by-environment interactions in calling properties, we provided evidence that male attractiveness was consistent across different thermic conditions (first prediction). Moreover, in 1 breeding season, we showed a statistically significant association between mating success and call rate, call duration, and spectral properties, suggesting that multivariate directional selection did act on these properties (second prediction). However, we found no conclusive evidence for either accepting or rejecting our third prediction. Indeed, although paternal call rate showed a marginally significant, negative correlation with offspring age at metamorphosis, the other call properties did not show any statistically significant association with offspring fitness. These results suggest that, in our study population, alternative mechanisms, such as male–male competition for signal detectability, might be a stronger driving force for the evolution of male costly displays than good genes.