Though bicycle as a mode of transport has many environmental and societal benefits as well as health benefits, bicyclists are one of the most vulnerable road users. According to the report by the Ministry of Road Transport and Highways (MoRTH, 2017), there is a sharp increase in the number of fatal victims in respect of bicyclists in 2017 over 2016. The number of cyclists killed jumped from 2585 in 2016 to 3559 in 2017, a 37.7% increase. In the present study, we present the analysis of the effect of the crash, geometric, environmental and cyclist characteristics on the bicycle–vehicle involved collisions by using the crash dataset of nine years (2009–2017) from Tamilnadu RADMS (Road Accident Data Management System) database with the application of fast and frugal tree (FFT) heuristic algorithm. The complete dataset (9978 crashes) was divided into two separate datasets: training data (6984 crashes) for the development of model and testing data (2984 crashes) for the performance evaluation. FFT algorithm identifies five major hues or variable attributes that influence the severity of bicycle crashes. The five major hues include the number of lanes, road separation, intersection, colliding vehicle type and road category. From the results of the present study, FFT acts as a complementary tool to other complex machine learning algorithms such as support vector machines, random forest, logistic regression and CART. The findings of the present study provide important insights for reducing the severity of bicycle-involved crashes at the planning and operations levels. © 2020 Informa UK Limited, trading as Taylor & Francis Group.