Real-time queue length estimation and prediction provides useful information for proactively managing transportation networks. Queue spillback from off-ramps onto main freeway lanes is a serious traffic issue that can be efficiently managed using dynamic queue information. In this thesis, a case-based reasoning algorithm combined with a Kalman filter is developed to provide real-time queue length measurements and predictions on long freeway off-ramps. Estimations are based on occupancy readings from three loop detectors installed on the ramp. The proposed method is examined using a micro-simulation model in a Quadstone Paramics package on an off-ramp‏ with a length of 650 meters. The simulation results demonstrate that the model is capable of estimating and predicting the queue length on long off-ramps in 60 second time intervals. The performance of the algorithm is examined under various demand loading scenarios, estimation time intervals and number of detectors through several experiments.