We study the extent of persistence in unemployment and low-pay employment in Italy in the period 2014-2017, using the Italian component of the EU-SILC survey merged with administrative data. We model persistence in unemployment and low-pay employment using different dynamic random-effects models accounting for observed and latent individual heterogeneity as well as endogeneity of the initial conditions. We find evidence of true state dependence in place for both unemployment and low-pay employment. Moreover, past unemployment spells increase the probability of being low-paid, conditional on being employed, while the opposite effect is limited. For both processes the degree of reliance on the previous state is considerably greater than the magnitude of cross-effects. Thus, evidence is presented that these processes shape almost independent no-pay/low-pay tunnels leading individuals into two different traps, rather than a cycle between the two states. •We study persistence in unemployment and low-pay in Italy using linked survey-administrative data.•Evidence is presented of state dependence for unemployment and low-pay employment.•Independent no-pay/low-pay tunnels lead individuals into two different traps.•The structure of the Italian labour market resembles more to a tunnel than a cycle.