The figures for out-of-school children put out by various official sources in India show wide variations. The problems lie not just in the definitions but also in methods of estimation. A glaring lacuna in this process is that sporadic or irregular attendance is not taken into account when estimating the number. This paper unpacks the phenomenon through an intensive micro-study of enrolment and attendance of all children in a single panchayat in India. It shows that irregular attendance accounts for a much larger proportion of out-of-school children, with wide variation in attendance across social groups. It also conducts a regression analysis to analyse school and household-level factors that affect student attendance. It finds that school-level factors play a much larger role in determining student attendance.