This paper examines the signaling role of workers’ parental leave choices theoretically and empirically. Human capital theory predicts that wages should depend on the absolute duration of time out of work. However, using administrative data from Denmark, I find that women who take a given amount of leave earn higher wages after returning to work if other mothers, in the same childbirth cohort, take longer leaves. The importance of relative leave duration suggests the possibility that leave duration choices signal labor market preferences. I develop a model which posits that firms infer private information about commitment to work through their choice of forgoing paid leave to return to work early. The model delivers distinct predictions of the signaling channel of parental leave when there is an exogenous change in the maximum allowed paid leave duration. I test these predictions using unanticipated leave extension policies in Denmark. I show that changing the equilibrium has indirect effects in changing the signals that are sent by all workers, and also has direct effects in terms of labor market consequences. Consistent with the model’s predictions, a leave extension causes infra-marginal mothers, whose leave would not have been constrained by the previously lower maximum, to take longer leave. For mothers for whom the previously lower maximum would have been binding, signaling contributes to a divergence in wages due to the information that their choices convey upon a leave extension. The paper provides direct evidence of persistent labor market consequences of signaling in a context in which signaling operates during the course of one’s labor market experience.