This paper presents the first longitudinal estimates of the effect of work-related training on labour market outcomes in Switzerland. Using a novel dataset that links official census data on adult education to longitudinal register data on labour market outcomes, we apply a regression-adjusted matched difference-in-differences approach with entropy balancing to account for selection bias and sorting on gains. We find that training participation increases yearly earnings and reduces the risk of unemployment already two years after the treatment. However, the effects are heterogeneous as to gender, age, education, and regional labour market context. Gains are highest for middle aged men with formal vocational education working in either depressed or booming labour markets.