Relying on private tutoring expenditures in South Korea, this paper examines whether large family size has a strong negative impact on educational investments in children. To deal with endogeneity of family size, the paper employs a non-parametric bounding method as well as an instrumental variables method. Our primary finding is that quantity–quality trade-offs in educational investments function in a way that varies by the sex of the child. While there is a non-negligible negative effect of large family size on educational investments for girls, there is little if any impact on those for boys. Son preferences traditionally shown by Korean parents seem to underlie such empirical findings.