<p>We present a Bisexual Galton-Watson process to model the dynamics of populations with sexual reproduction, allowing overlaps in different generations. We study the time to extinction of this model, for which we provide a sufficient condition that ensures that extinction occurs with probability 1. This condition extends classical results from the literature of branching processes. Furthermore, we show that the time to extinction of our model can be approximated using standard tools from Extreme Value Theory, for which we make use of a methodology based on simulations of the time to extinction. This methodology is presented in full detail and is later applied to three numerical examples.</p>