Electrokinetic models for electrolyte transport in narrow sinusoidal capillary tubes were used to study salt rejection. Two approaches were adopted. The first formulation made use of Gouy-Chapman model of the double layer coupled with the creeping flow and Nernst-Planck equations. The elliptic nature of the formulation was retained. The second formulation was an adaptation of Dresner’s model where Boltzmann distribution was assumed to hold in the radial direction. The salt rejection as evaluated by the two formulations was fairly close. However, the details of the electrokinetic field variables obtained from the two formulations were not in close agreement. The dimensionless inverse Debye length, ka, based on the average tube radius, a, was found to play a major role in the salt rejection. When the electrical double layers overlap, i.e., ka is small, the salt rejection is governed by the surface potential, flow Peclet number, and the amplitude of the tube surface. In addition, for high Peclet numbers, the tube surface potential becomes the sole parameter controlling the salt rejection. When the electrical double layers do not overlap, i.e., ka is large, the salt rejection becomes a strong function of the tube geometry as well as the surface potential and Peclet number.