To lower the cost of a force-reflecting teleoperation system, position errors instead of direct measurements from force/torque sensors can be used for force feedback. An investigation on transparency shows that position-error-based teleoperation provides poor transparency when using constant controller gains, and this can be overcome by gain switching. The scenario studied in this paper is a position-error-based bilateral teleoperation system when the slave robot is either in free motion or colliding with a stationary stiff environment. A contact stability analysis is presented for teleoperation systems with constant controller gains and then extended to a gain-switching teleoperation system. Issues on the design of a gain-switching control scheme are discussed. The effectiveness of the proposed gain-switching control scheme is evaluated experimentally.