In the empirical literature on child labour and income transfer programmes, evidence has been lacking on whether (and how) the size of transfer influences the impact on child labour. This paper finds significant size effects in the impact of scholarship-based transfers on child labour in Nepal. High-value scholarships decrease 8–16 year-old girls’ total work hours by one-third, largely reducing their hours in paid and unpaid economic activities with little impact on domestic chores. Low-value scholarships have no impact at all. The findings elucidate the scope for calibrating transfer size to achieve greater child labour impacts in developing countries.