We estimate the effect of motherhood on wages using matching. We distinguish between net and direct effects. The net effect includes the total wage costs, whereas the direct represents the causal effect. Since covariates are likely affected by motherhood, the latter effect is not immediately uncovered. We therefore implement two strategies: first, we confine the analysis to consider sector-specific treatment effects; second, we impose additive separability on the outcome equation. We find negative net effects that vary little with sector. The direct effect is small and negative in the public sector and insignificant in the private sector.