This paper uses data from Peru, Pakistan and Ghana to analyse simultaneously child labour and child schooling, and compares them between these countries. We use a multinomial logit estimation procedure that analyses the participation and non-participation of children in schooling and in employment and, in particular, allows the possibility that a child combines schooling with employment or does neither. We also use an ordered probit estimation procedure based on a ranking of the various child schooling/employment/non-schooling/non-employment outcomes. The results point to both similarities and striking dissimilarities in the nature of child labour and child schooling between the chosen countries. For example, in Pakistan, but not in Peru, the girl child's ordering of schooling/employment outcomes shows her at a position of extreme disadvantage. Household poverty discourages a child from achieving superior outcomes, but the effect varies markedly across the three countries.