Train car crowding is a key issue resulting in passenger dissatisfaction and degraded transit service. One reason for train car crowding is an uneven distribution of passengers among the train cars which results in overcrowding of certain cars while leaving other cars sparsely occupied. If passengers waiting at a station are informed about the expected crowding levels in the arriving train, they can position themselves on the platform to board a less-crowded car. ARCON proposes a crowding information collection and dissemination (CICD) system for predicting passenger loading on each train car and providing the crowding information to the passengers. The CICD system is a cost effective solution that comprises of three subsystems: Automatic Passenger Counting (APC) system, Passenger Flow Prediction (PFP) system, and Passenger Information System (PIS). The APC subsystem uses an accelerometer data-based algorithm together with currently available passenger count technologies to measure passenger loading in each car. The PFP subsystem uses a statistical model of passenger flow to predict the number of passengers that will alight and board the train. The PFP subsystem updates and refines the APC measurements. Finally, the PIS subsystem leverages existing data links and infrastructure to disseminate the crowding information to the passengers.