Background A significant number of pedestrians are willing to take risk while crossing the road at all type of crosswalks. The paper explored the utility of binary probabilistic modelling and its performance measures to analyse the pedestrians’ unsafe crossing behaviour exhibited by different types of pedestrians and their associated risk at different traffic and road environments.

Methods Binary Logistic regression models were fitted to determine the probability of road crossing by a pedestrian with a set of predictor variables. To quantify the risk to pedestrian, adequate gap size to cross the road at the crosswalks was determined. Thereafter, binary Logistic regression models were fitted to determine the probability of crossing by a pedestrian with the gap size less than the adequate gap size for crossing.

Results Findings reveal that crossing behaviour of pedestrian significantly varies at different crossing locations. At signalised intersections and a free left turn crosswalk, no other predictor parameter, except the gap size parameter, is contributing to determine pedestrians’ crossing behaviour, whereas at non-signalised crosswalks (at the foot of flyover and grade-separators and non-signalised crosswalk near bus-stop), their individual and traffic characteristics also contribute to determine their unsafe road crossing behaviour. At all types of crosswalks, almost all the pedestrians (who cross the road in an unsafe condition) cross with the gap size less than the adequate gap size.

Conclusions The results of the study are the basic inputs to understand the road crossing behaviour of pedestrian and their associated risk needed to design well structured and safe transport system. The results highlight human behaviour and risk taking owing to road geometry and traffic operations. Free left-turns, flyovers or grade separators are often introduced to reduce traffic congestion. However, the study shows the negative impact of these types of constructions on the pedestrians.