Abstract Despite the large body of ALMP evaluations focussing on isolated training programmes for unemployed jobseekers, our understanding of potential reasons for (in)effectiveness remains limited. Specific training programmes aim to remediate particular supply- or demand-side barriers to employment experienced by targeted jobseekers. Consequently, this study unpacks training into four different types: (I) general classroom training (GCT) to enhance motivation and job search skills, (II) occupation-specific classroom training (OCT) addressing gaps in human capital, (III) non-contractual workplace training (NCWT) combining human capital acquisition with workplace experience, and (IV) contractual workplace training (CWT) additionally including a temporary wage subsidy to reduce hiring costs for employers. Using large-scale longitudinal register data, dynamic propensity score matching, and hazard models indicate positive effects of OCT participation, and particularly NCWT programmes allowing human and social capital accumulation in a workplace setting, on the transition into (stable) regular employment. In contrast, the non-effects for GCT participants highlight the need for more follow-up programmes, and the fact that, after controlling for the selective recruitment by employers of unemployed jobseekers with relatively strong profiles, CWT programme participants show moderate, short-lived positive effects which might inspire policymakers to reconsider programme assignment in light of cream-skimming by employers.