This study examines the relationship between fertility decline and the reversal/narrowing of the gender gap in literacy/numeracy test scores. Drawing on Becker's Quantity-Quality (Q-Q) trade-off model, we propose that in a society such as China, where son-preference is prevalent, the Q-Q trade-off would be larger for daughters than that for sons. An exogenous reduction of fertility would make girls more likely to live in a single-sex family, which in turn increases the share of human capital investment for girls. We test this empirically. To consider the endogenous nature of the demand for children, we exploit an exogenous variation in fertility due to China's family planning policy. Utilising the policy intensity information collected from hundreds of county gazetteers to construct a novel instrument for fertility, we find that a reduction in one sibling narrows the gender gap in numeracy and literacy test scores by 14.8% and 21.4% of a standard deviation in the rural sample and 4.0% and 6.5% in the urban sample. The pattern is more pronounced in regions with a higher proportion of people who prefer a son over a daughter. We also provide suggestive evidence that the channel of the effect is indeed largely through the increased probability of girls living in single-sex families.