This study uses Nepalese data to estimate the impact of individual, household andcluster/community level variables on child labour and child schooling. The principalestimates are, then, compared with those from Bangladesh and Pakistan. The exercise isdesigned to identify effective policy instruments that could influence child labour andchild schooling in South Asia. The results show that the impact of a variable on achild’s education/employment is, often, highly sensitive to the specification in theestimation and to the country considered. There are, however some results that are fairlyrobust. For example, in both Nepal and Pakistan, inequality has a strong U shapedimpact on both child labour participation rates and child labour hours, thus, pointing tohigh inequality as a significant cause of child labour. In contrast, household poverty hasonly a weak link with child labour, though it seems to be more important in the contextof child schooling. The current school attendance by a child has a large, negative impacton her labour hours, thus, pointing to compulsory schooling as an effective instrumentin reducing child labour. Other potentially useful instruments include adult educationlevels, improvements in the schooling infrastructure, and the provision of amenitiessuch as water and electricity in the villages.