Research on the effect of in utero shocks on health at birth may be influenced by in utero selection. This study outlines a conceptual framework and shows that the results of the standard empirical approach are biased if (i) the exposure changes the probability of fetal death and (ii) health differences exist between deceased and surviving fetuses. Furthermore, an empirical example is provided to illustrate, the potential importance of fetal selection. Examining the impact of heat on birth weight, I find that accounting for fetal selection substantially increases the heat effect compared to the standard approach. These results suggest that incorporating the distorting effect of fetal losses into the estimations may be critical in some cases to provide more informed guidance for public policy.