Quarterly collections of demersal fish were conducted between 2014 and 2018 on the adjacent continental shelf and in the Paranaguá Estuarine Complex (PEC). Univariate and multivariate statistical analyses were used to determine the abundance, richness, and composition of demersal fish assemblages and to list predictor variables responsible for the variability in these ecological descriptors by applying generalized linear models. The collected specimens of demersal fish (n = 25,179) were distributed across 19 orders, 31 families, and 72 species. Fish richness was explained by the following predictor variables: the percentage of fine sand in the sediment, the richness of the benthic megafauna, and sampling sites. The abundance was associated with the predictor variables of the percentage of very fine sand in the sediment, percentage of silt in the sediment, salinity, and sampling sites. Canonical correspondence analysis showed that the fish assemblage composition was strongly associated with the salinity gradient and positively correlated with benthic megafaunal species richness. As it is one of the last remnants of the Atlantic Forest, declared by UNESCO as a Natural Heritage of Humanity, the patterns observed in this study will support PEC's coastal management of current developmental demands, particularly related to port activities.