The objective of this study was to analyze traffic accidents involving motorcycles in Belo Horizonte, Minas Gerais State, Brazil, from 2007 to 2011 and to identify clusters of high-risk and hazardous intersections in and around the city. Data were provided by the Military Police Brigade and the Emergency Medical Service (SAMU). Accident severity rates were used to identify critical intersections. Two techniques were used: kernel analysis and scan statistics (continuous Poisson model). High-risk clusters were located in the downtown area and on major thoroughfares. Surprisingly, the highest risk of accidents and death occurred not at intersections, but between them. Hazardous intersections are part of routes used to access regions around Greater Metropolitan Belo Horizonte. Two distinct trends in mortality rates and accidents were identified. Most motorcycle deaths occurred after 7:00 PM. The study concludes that there is an urgent need to improve motorcycle and public transportation routes.