We propose a novel son preference measure that relates the preference to a specific child. We find child-specific son preference to be more common among later born children and in families with fewer sons. Using the novel measure and an interaction instrumental variables approach, we estimate a penalty in early mental functions for unwanted girls of 0.7 standard deviations. This penalty appears to be partially driven by discrimination against girls and partially by pampering of boys. Children’s health and parental inputs do not mediate the effect from son preference to mental development. Our findings highlight the relevance of parents’ attitudes for a nurturing home environment and healthy brain development.