Current statistics on child labor are generally based on economically active children. This paper will argue that these figures are not a workable proxy for data on child labor, generating numbers of child laborers and their gender composition that do not represent the group described by the international definition of child labor. This raises the question of reliable alternative ways of measuring children's activities with the aim of analyzing the incidence of child labor. The paper addresses this and proposes a child labor module that can be linked to surveys of labor force or living conditions. It also proposes some ideas for how to analyze data on children's activities and child labor.