Primary enrolment rates are very high in Peru, but so are the failure and drop-out rates. Thus an understanding of the nature of child schooling should consider school progression from primary to secondary and higher levels, taking account of the conditional sequence with the previous level and self-selection into the next higher level of schooling. Using a unique correlated sequential probit model with unobserved heterogeneity the present paper does so and obtains richer results, argued to be better than the standard static estimates. It is shown that the same set of individual/parental/household characteristics may affect different levels of schooling differently.