Resumo. Abstract Aim β diversity was used as an indicator to test the hypothesis that a fishway can increase river connectivity, as a reduction in the value of this metric indicates greater similarity among fish assemblages. Methods Quarterly sampling was performed with gillnets upstream and downstream of the Porto Primavera dam, Upper Paraná Basin, before the opening of the fishway between February 1999 and May 2001, and between February 2003 and May 2005 with the fishway in operation (10 samples per period/site). The correlation between the fish assemblages and sites was verified by applying two Mantel tests (downstream versus upstream before and one equivalent after the fishway) using the distance matrices generated by the Bray‐Curtis index. The β diversity between downstream and upstream was calculated using the Sørensen index for each sample for the comparison of such values before and after the operation of the fishway, which were tested with Asymptotic Wilcoxon-Mann-Whitney Test. To summarize changes in the structure of the fish assemblages between the sites and periods nonmetric multidimensional scaling (NMDS) was applied based on Bray-Curtis dissimilarities, using multi-response permutation procedures (MRPP). Results The Mantel tests showed no correlation between the upstream and downstream assemblages before the operation of the fishway but greater correlation after. β diversity was lower after than before the fishway, representing an increase in similarity between downstream and upstream with the fishway. MRPP analysis indicated that the fish assemblages between the sites differed more than was expected by chance, with smaller divergence between the sampling sites after the fishway (A = 0.08) than before it opened (A = 0.09). Conclusions Although fishways remain an incipient and controversial theme for neotropical fish, the Porto Primavera fishway appears to be effective in terms of the restoration of connectivity.