The ability to monitor vehicle queue length accurately at metered on-ramps can improve ramp meter performance and help to create improved ramp metering algorithms. The queue length distribution can be considered as a continuous analog signal, which consists of both deterministic and stochastic components. Three types of methods for the estimation of on-ramp queue length are discussed: Kalman filter, linear occupancy, and "Highway Capacity Manual" (HCM) back of queue. Queue data estimated with these methods are compared with field-observed queue data and random number samples. The comparisons indicate that the Kalman filter and linear occupancy methods are usable for real-world operations, but both of them have limitations. The HCM back-of-queue method does not produce reliable estimates for on-ramp queue length.