This paper employs a hurdle model approach to ask whether the extent of gender bias in education expenditure within rural households in India changed over time from 1995 to 2017-18. Our most striking finding is that there has been a change over time in the way that gender bias is practiced within the household. In 1995, gender bias occurred through a significantly higher probability of school-enrolment of boys than girls, but by 2017-18, gender bias was practiced via significantly higher conditional education expenditure on boys than girls, and this was largely achieved via pro-male private school enrolment decisions. Households practicing gender equality in school enrolment by 2017-18 is a desirable trend. However, girls' significant disadvantage vis-à-vis boys in terms of lower education expenditure, achieved via their lower private school enrolment rate by 2017-18, is problematic if lower expenditure is associated with lower levels of cognitive skills (literacy, numeracy, etc.) since both individual economic returns and national economic growth accrue to cognitive skills and not independently to completing a given number of years in school. Household fixed effects analysis shows that the observed gender biases are a within-household phenomenon rather than an artefact of differences in unobservables across households.