We analyse the consequences of starting a wage subsidised job, “marginal employment”, for unemployed workers. Marginal employment is a type of wage subsidy paid to unemployed workers and they do not lose their unemployment benefits if the wage is below a certain threshold. We ask if the unemployed who start marginal jobs face better labour market outcomes than those who do not work. A priori it is not clear if those who work in marginal employment improve their labour market status, e.g. by signalling effort, or worsen it by reduced job search effort. We selected unemployed workers and investigate the effect of marginal employment on their labour market outcomes, by means of propensity score matching. Our results suggest that selection into marginal employment is “negative”, i.e. workers with characteristics we usually associate with low-productivity are more likely to select into such jobs. The unemployed who start to work in marginal employment during their unemployment spell suffer a (causal) penalty for doing so, relative to their peers who do not. The penalty, in terms of less employment, more unemployment, lower wages, lessens over time but is still present after three years.