Repeated association between subsets of individuals is a common feature of species living in social groups. Because colonial breeding, an extreme case of group living, is associated with certain group behaviour, colonial species are interesting study models to explore the occurrence of social bonds between individuals. As colonial species are usually highly philopatric, disentangling the fidelity to a breeding site from the fidelity to a group of individuals is challenging. Slender-billed gull, Chroicocephalus genei, colonies, however, relocate almost yearly. This behaviour makes it possible to study individual associations over several years, i.e. associations between individuals breeding in the same colony in more than 1 year. To quantify and identify the mechanisms that favour repeated individual associations across years, we analysed data from 14 years of observations of 953 individually marked gulls. Our results showed that some individuals repeatedly bred together across breeding seasons despite the colony moving every year. The probability of an individual selecting a colony increased with an increase in the number of birds that had bred in the same colony as that individual the previous year and not the overall number at the colony. However, we found yearly variation in group tenacity levels and that colony breeding failure favoured splitting of the groups. We also found that association rates rapidly decreased across years but stayed higher than random associations during 2 consecutive years after the first observations. Moreover, over the entire study period, we plotted a bipartite network and found that all colonies and individuals in the population were fully connected. This study reveals group tenacity across years in a colonial bird. Evolutionary pressures as well as the mechanisms favouring social bond persistence across years in colonial birds, however, need further research.