Many countries impose temporary employment bans to recently arrived asylum seekers, a period during which they are not allowed to take up legal employment. In this paper we assess the long-term impact of such temporary employment bans on refugees’ labour market integration. To do so we use micro-data from the European Labour Force survey that allow to identify migrants who arrived in Europe seeking humanitarian protection and combine them with newly collected data on the presence and length of employment bans across European countries, from 1985 onward. We estimate DID regressions exploiting within-country variation in employment bans and find that being exposed to an employment ban at arrival reduces employment probability in the medium-run by 15%. Importantly, these effects are not mechanic, persist over time – fading away only after about 15 years – and are non-linear in the length of ban duration. The negative impact on employment primarily comes from reduced participation, suggesting that employment bans may persistently push refugees out of the labour market and into welfare. Our causal estimates are consistent across other two identification strategies, a triple DID where other migrants are used as control group, and an IV strategy that exploits a 2003 EU directive that set a maximum duration of 12 months for any employment ban.