Samples of soil and recently deposited sediments were collected from the river bed, bank face and bank tops of two lengths (reaches) of the River Frome, Dorset, UK and one reach of the River Tern, Shropshire, UK. Soil propagule bank samples were collected in May 2003, and depositional samples were collected subsequently over four consecutive 4-month periods between June 2003 and October 2004. The samples were subjected to three emergence trials under drained, waterlogged and submerged conditions. Significantly more seedlings germinated in the drained than waterlogged trial, and waterlogged than submerged trials. Drained, waterlogged and submerged trials identified 186, 76 and 37 species, respectively. Six species identified in the waterlogged trials were not identified in the drained trials, and five species in the submerged trials were not found in the drained trials. Submerged trials added two species to the drained and waterlogged results. Application of detrended correspondence analysis (DCA) to average species abundance data, for the sampled hydrological habitats (bed, bank face, bank top) within the soil propagule bank and four depositional data sets, generated different results for the three trials. The drained treatment revealed significant differences between sites, seasons and hydrological habitats, whereas the waterlogged and submerged treatments presented an increasingly homogeneous view of the samples. Our results confirm other propagule bank emergence comparisons and extend them to depositional samples, demonstrating that the strong environmental sieves imposed by waterlogging and submergence restrict emergence of numerous terrestrial, wetland and even some aquatic species that were successfully identified using the drained conditions.