We set up an on-the-job learning model to explain how spare-time on-the-job training affects life skills formation of adolescents. We obtain causal estimates by exploiting variation between employed twins and estimation of a value-added model using full population Danish administrative data. We find that spare-time work experience has no effect on school absenteeism, reduces crime, increases school grades and the speed of enrollment into upper secondary education. We interpret our findings as evidence that the positive effect from on-the-job learning in spare-time jobs on life skill formation more than outweigh the potential negative consumption and time-use externalities of spare-time work.