The estimation of vehicle lateral speed, a critical variable for vehicle stability control, four-wheel-steering and other advanced dynamic control systems, is studied in this paper. We presented three different approaches, one each from three categories: transfer function approach, state-space approach, and kinematics approach. The first two methods rely on a vehicle dynamic (bicycle) model, and the last approach is based on the kinematics relationship of measured signals. The basic formulation of all three methods assumed that the road bank angle is negligible, and thus needs to be enhanced by a road bank angle estimation algorithm to work satisfactorily when the road bank is significant. The performance of these three (enhanced) methods are investigated using simulation and experimental data. For the experimental verification, we present four cases: nominal (high friction, flat road), banked road, low-friction, and low-friction-near-spin. Weakness of the three estimation algorithms is discussed.