We estimate the impact of social assistance on youth employment in Denmark, 2000-06. For childless unmarried individuals, maximum welfare payments increased by 55% at age 25. Using administrative panel data, we find a significant disincentive effect among the low-skilled only, for whom employment fell by 2%-3% and benefit take-up increased by 10%-14%. Two-thirds of the effect is attributable to transitions from work to welfare and one third to reduced labor market entry. Heterogeneous estimates along the earnings distribution showing that employment responses are concentrated at very low earnings. We find no evidence of an age discontinuity in criminal activities.