After the 1990s recession in Sweden, social assistance (SA) recipiency decreased to levels that were clearly lower than before the recession. However, this decrease masked a strong polarisation: the number of short-term recipients fell, but the number of long-term recipients was higher than before the recession. This article shows how SA recipiency and dependence changed over a whole economic cycle in Sweden's largest city, Stockholm, and asks whether the increasing dependence can be explained by immigration. It is shown that the relative increase of long-term SA is similar among immigrants and native-born, but decomposition analysis reveals that the increase among native-born is of minor importance for the overall increase. Nearly half the increase can be attributed instead to the increased representation of immigrants in the population, and another 38 per cent to increased dependence among immigrants. Only 15 per cent of the total increase in long-term SA is a result of increased dependence among native-born.