Hydrological connectivity is one of the main controlling factors of habitats and aquatic assemblages on river floodplains. Nevertheless, the lack of universal measures of river–floodplain connectivity (i.e. the lateral hydrological connectivity, LHC) limits the comparison of the response of aquatic assemblages to hydrological connectivity and impedes the understanding of floodplain functioning across different systems. To address these needs, we tested the ability of six different LHC surrogates to model changes in richness, abundance and composition of aquatic assemblages across a Mediterranean floodplain (Ebro River, NE Spain). As shown by generalised additive models, LHC surrogates explained 15% to 65% of the richness and abundance of aquatic assemblages. Zooplankton, macroinvertebrates and phytoplankton showed overlapping peaks of richness at flood duration rates of 5, 15 and 30 days year^–1 respectively. Redundancy analyses showed that LHC surrogates explained 17% to 37% of aquatic assemblage composition. Distance to the river and flood duration were the most important determinants of macroinvertebrate composition, whereas flood magnitude and water-level variability best accounted for the variance in zooplankton and phytoplankton compositions. Models based on LHC surrogates such as those presented here can help in predicting the consequences of restoration measures and may be useful in setting restoration goals for aquatic assemblages.