Community development in restored streams is often slow or even absent, but reasons remain obscure. Inadequate restoration measures, catchment-scale pressures, community closure and colonization barriers all may prevent or slow down recovery processes. For initial colonization, dispersal processes are supposed to be most important, which are referred to as dispersal filter. Environmental conditions of a restored reach determine if a dispersing species can successfully establish (environmental filter). Lastly, while available niches at those reaches fill up, biotic interactions, such as competition, become more important (biotic filter). To investigate the importance of these different filters, we compared benthic invertebrate communities of 20 sites in the Boye catchment (Western Germany), a former open sewer system. The sites were grouped, based on the years since restoration, into ‘unimpacted’ (never restored), ‘recently restored’ (< 4 years) and ‘mature restored’ (> 10 years) sites. Data collected at 28 additional sites in the catchment informed us on distances to potential source populations. Habitat suitability describes the fit between environmental conditions (abiotic site data) and species preferences and was used to assess the role of environmental filtering. We evaluated the role of the biotic filter based on trait overlap, referring to possible interspecific competition.Communities collected at recently restored sites differed from those of mature restored and unimpacted sites. Taxa present at recently restored and mature sites had closer source populations than those of unimpacted sites. Taxa at mature and unimpacted sites had a better fit to the present habitats than those of recently restored sites. The trait overlap did not differ between co-occurring and not co-occurring taxa at any of the site groups. Our findings show that communities of mature restored sites that have been restored more than 10 years ago, resembled those of unimpacted sites. Dispersal was most important in early years of recovery. Taxa occurrences at sites with nearby source populations and low habitat suitability are likely the result of high rates of dispersal from upstream sources (mass effects). These can be caused by hatching events or environmental disturbances. Habitat suitability played a larger role for communities at mature and unimpacted sites which indicates that optimal communities shape over time. We did not find indications that competition played a role for community assembly. Hence, dispersal and habitat suitability were most relevant for species’ occurrences. Competition could be more important on micro scales and the results may differ if species abundances are taken into account.Graphical Abstract