In the modelling of sea ice motion, a major concern focuses on how sensitive model simulation results are to changes in the model inputs or the system parameters. Parallel to this concern is the manner in which the uncertainty in the model inputs or system parameters manifests itself as uncertainty in model results. In this paper, the theory required for the sensitivity and uncertainty analysis of a short term ice motion model is presented. The governing equations for this ice motion model are composed of the standard equation of motion in conjunction with a two-level thickness distribution model. To ascertain model sensitivities, a method based on the adjoint operator technique is utilized. Uncertainties in the model simulation results are determined by a first-order, second-moment method. Two formulations of the adjoint state equations are presented, one involving the governing partial differential equations (continuous formulation) and the other involving the discrete or numerical form of the governing equation (discrete formulation). The resulting sensitivity and uncertainty methodology is applied to a 4-day ice motion episode in the southern Beaufort Sea. The sensitivity and uncertainty of the calculated ice velocity and the ice thickness at a particular location within the domain are discussed. ¿ American Geophysical Union 1990