Drawing on the nationally representative "Participation and Expenditure in Education" surveys, we document the incidence and cost of private tutoring at different stages of schooling over the last two decades in India. As private tutoring involve two decisions: a) whether to take private tuition or not, and b) how much to spend on private tutoring conditional on positive decision in (a), we analyze the determinants of the two decisions separately using a Hurdle model. We find that private tutoring is not a new phenomenon in India: a significant proportion of students at each stage of schooling took private tutoring even in 1986-87, and there has been no dramatic increase in those proportions. Students in urban areas and private schools are not only more likely to take private tutoring but also spend more on private tutoring. We also find that demand for private tutoring is inelastic at each stage of schooling, which implies that the private tutoring is a necessary good in the household consumption basket. We also find evidence of pro-male bias in both decisions regarding private tutoring.