This paper studies the relationship between single motherhood and children's height-for-age z-scores in Brazil. In order to isolate the causal effect between family structure and children's condition, we estimate an econometric model that uses male preference for firstborn sons and local sex ratios to instrument the probability of a woman becoming a single mother. Our results have a local average treatment effect interpretation (LATE). We find that children being raised by a single mother (whose marital status is affected by a firstborn girl and a low sex ratio) have a height-for-age z-score that is lower than that of children of similar characteristics that cohabit with both progenitors. We claim that the increasing trend of single motherhood in Brazil should be of concern in health policy design.