In this paper, the authors provide the first comprehensive analysis of the labour market performance of refugees vis-à-vis their host community members. We consider the case of Uganda, which has a progressive legal environment for refugees, including the right to work, freedom of movement and the right to cultivate land. We use comprehensive pre-COVID data supplemented by existing data during and after COVID-19 to track the labour market performance of refugees vis-à-vis their host community members. The results indicate that refugees in Uganda are 31 percentage points less likely to be employed than hosts. Using a supervised machine learning algorithm, we show that a small set of observable traits, such as age, gender, employment status before displacement, education, local labour markets, and social network, predict the current employment status of refugees. While refugees face understandable challenges in participating in the labour market upon arrival, it is striking that the gap persists over time, sug