This dataset contains a collection of 30 datasets on geographically located cycling crashes across the world. This collection covers country, region, or city-wide areas in multiple easy-to-use formats, such as Comma Separated Values (CSV) files and GeoJSON files. Crashes and accidents are often published by police, cities, or other governmental agencies, but are often difficult to use and contain many missing values or errors. This collection results from the retrieval of each dataset from such sources and its subsequent cleaning, curation, and validation. The collection includes data from: Barcelona, Cambridgeshire, Chicago, Colorado, Connecticut, Denver, Detroit, France, Geneva, Germany, Helsinki, Las Vegas, Los Angeles, Louisville, Madrid, Nantes, Nashville, Netherlands, New York, Pasadena, Pennsylvania, Queensland, Richmond, Rome, San Jose, Seattle, United Kingdom (Collideoscope), United Kingdom (Gov), Victoria, Washington DC. A detailed documentation of the data will be available once the data descriptor manuscript has been finalized and published.

The licensing terms are documented separately for each dataset. This work was partially funded by Funda����o para a Ci��ncia e Tecnologia via grant [PD/BD/142948/2018], and LARSyS---FCT Project [UIDB/50009/2020].