We evaluate a Danish reform focused on improving Danish language training for those granted refugee status on or after January 1, 1999. Using a Regression Discontinuity Design we find a significant, permanent, positive effect on earnings. This effect emerged after completion of language classes and was accompanied by additional schooling and higher probability of working in communication-intensive jobs, suggesting that language training, rather than other minor aspects of the reform, produced it. We also find evidence of higher completion rates of lower secondary school and lower probability of crime for male children with both parents exposed to the reform.