Recent work increasingly reveals the importance of social information in individual dispersal decisions, population dynamics and conservation. Much of the knowledge gained to date comes from studies on short-lived and/or densely breeding species. In contrast, our understanding of the processes involved in nest-site selection for long-lived, solitary breeding species is insufficient. We increased nest-site availability by nest-box supplementation over a 5-year period in a population of a long-lived, solitary, secondary-cavity nesting bird, the European roller Coracias garrulus, breeding in natural cavities and human constructions. We tested the nest limitation and the inadvertent conspecific social information hypothesis in order to study the dynamics and mechanisms of abandonment of previously used nests and the colonisation of new ones. Our data lend support to the nest-limitation hypothesis both in terms of quantity-population and the size of breeding clusters increased, and suitability--the majority of pairs used and re-occupied nest-boxes. Nevertheless, the use of natural cavities did not decrease after 5 years. At the between-patch scale, rollers were revealed to colonise nest-boxes based on conspecific social attraction, namely distance to the nearest neighbour in the same season. Despite the unpredictability of patch productivity, at the within--patch scale, the selection of previously unoccupied cavities was consistent with the performance-based conspecific attraction hypothesis. Philopatry could account for the repeated use of cavities, because nests that were used for two successive years were more likely to also be reused in the subsequent season.