Child penalties in labour market outcomes are well-documented: after childbirth, mothers' employment and earnings drop persistently compared to fathers. In addition to gender norms, a potential driver could be the loss in labour market skills due to mothers' longer employment interruptions. This paper estimates child penalties in adult cognitive skills by adapting the pseudo-panel approach to a single cross-section of 29 countries in the PIAAC dataset. We find a persistent drop in numeracy skills after childbirth for both parents between 0.13 (short-run) and 0.16 standard deviations (long-run), but no statistically significant difference between mothers and fathers. Estimates of child penalties in skills strongly depend on controlling for pre-determined characteristics, especially education. Additionally, there is no evidence for worse occupational skill matches for mothers after childbirth. Our findings suggest that changes in general labour market skills cannot explain child penalties in labour market outcomes, and that a cross-sectional estimation of child penalties can be sensitive to characteristics of the outcome variable.