Willow removal followed by riparian revegetation is a widespread river restoration practice in Australia, but the ecological response to this has rarely been evaluated. We sampled river macroinvertebrates from six sites each of three riparian vegetation types: revegetated (treatment), willow-dominated (control) and native forest (reference) in the Gellibrand River catchment during austral spring 2007 and autumn 2008, and measured temperature and light intensity. Revegetated sites varied in age from 1 to 8 years since restoration. Abundances of invertebrates were similar across vegetation types, but were higher during autumn. Macroinvertebrate assemblages at revegetated sites (regardless of age) and at willow-dominated sites showed little among-site variation compared with native forest sites, which showed high site-to-site variability. Water temperatures and light intensity were higher at revegetated sites where works had recently been completed and cooler in native forest sites and long-established revegetated sites. The reduced variability in macroinvertebrate communities among revegetated sites may result from their history as willow-dominated sites or from the disturbance created by willow removal. Either way, these results suggest that longer than 8 years is required before macroinvertebrate assemblages in restored stretches of stream show the variation that appears characteristic of natural sites.