jats:titleAbstract</jats:title>jats:pRoad-based citizen science surveys are increasingly used for long-term monitoring of wildlife, including amphibians, over large spatial scales. However, how representative such data are when compared to the actual species distribution remains unclear. Spatial biases in site selection or road network coverage by volunteers could skew results towards more urbanised areas and consequently produce incorrect or partial trend estimations at regional or national scales. Our objective was to compare and verify potential spatial biases of road-based data using distribution datasets of different origins. We used as a case study the common toad (jats:italicBufo bufo</jats:italic>), a fast-declining species and the main amphibian targeted by conservation action on roads in Europe. We used Maxent models to compare road survey data obtained from the 35 year-long “Toads on Roads” project in Great Britain with models using national-scale toad distribution records as well as with models using randomly generated points on roads. Distribution models that used data collected by volunteers on roads produced similar results to those obtained from overall species distribution, indicating the lack of selection bias and high spatial coverage of volunteer-collected data on roads. Toads were generally absent from mountainous areas and, despite the high availability of potential recorders, showed nearly complete absence of road-based records in large urban areas. This is probably the first study that comparatively evaluates species distribution models created using datasets of different origin in order to verify the influence of potential spatial bias of data collected by volunteers on roads. Large-scale declines of widespread amphibians have been demonstrated using data collected on roads and our results indicate that such data are representative and certainly comparable to other existing datasets. We show that for countries with high road network coverage, such as Great Britain, road-based data collected by volunteers represent a robust dataset and a critical citizen science contribution to conservation.</jats:p>