Context. Citizen science is increasingly used to collect biodiversity data to inform conservation management, but its validity within urban greenspaces remains largely unresolved.Aims. To assess the validity of eBird data for generating biodiversity estimates within an urban greenspace.Methods. We compared data from structured avian surveys with eBird data at an urban greenspace in Sydney during 2012–16, using species richness and Shannon diversity indices. We also compared community composition, using non-metric multidimensional scaling (NMDS) and dissimilarities using non-parametric MANOVA.Key results. Structured surveys had a lower overall species richness (80 versus 116) and Shannon diversity (3.64 versus 3.94) than eBird data, but we found no significant differences when using years as replicates. After standardising the richness and diversity indices by time spent surveying in a given year, structured surveys produced significantly higher biodiversity estimates. Further, when grouped into species occupying different broad habitats, there were no significant differences in waterbird or landbird species richness, or in Shannon diversity between data sources.Conclusions. The most likely explanation for the larger magnitudes of the biodiversity indices from the eBird data is the increase in effort manifested in the number of observers, time spent surveying and spatial coverage. This resulted in increased detection of uncommon species, which in turn accounted for a significant difference (R2 = 0.21, P = 0.015) in overall community composition measured by the two methods.Implications. Our results highlight the opportunities provided by eBird data as a useful tool for land managers for monitoring avian communities in urban areas.