Artificial structures play a significant role as optional breeding places for the sand martin. However, it is virtually unknown which factors determine habitat selection in these artificial sites and to what extent the species selects the same key features that it would in its natural habitat. We compared 16 variables from artificial walls with pipes in channelled rivers between walls holding (N = 56) or not holding (N = 50) a sand martin breeding colony. The presence of colonies was negatively associated with the distance of the breeding wall from the water and positively associated with flooding risk. Bibliographic data were used as a reference for the characteristics of natural breeding habitat. The sand martin selected similar key factors in artificial habitats as when breeding in a natural setting. These factors could probably be linked to limiting predator access to the colony and the occurrence of periodical rinsing episodes that may minimise parasite loads in nesting cavities.