This paper considers feedback systems with asymmetric (i.e., non-odd functions) nonlinear actuators and sensors. While the stability of such systems can be investigated using the theory of absolute stability and its extensions, the current paper provides a method for their performance analysis, i.e., reference tracking and disturbance rejection. Similar to the case of symmetric nonlinearities considered in earlier work, the development is based on the method of stochastic linearisation (which is akin to the describing functions, but intended to study general properties of dynamics, rather than periodic regimes). Unlike the symmetric case, however, the nonlinearities considered here must be approximated not only by a quasilinear gain, but a quasilinear bias as well. This paper derives transcendental equations for the quasilinear gain and bias, provides necessary and sufficient conditions for existence of their solutions, and, using simulations, investigates the accuracy of these solutions as a tool for predicting the quality of reference tracking and disturbance rejection. The method developed is then applied to performance analysis of specific systems, and the effect of asymmetry on their behaviour is investigated. In addition, this method is used to justify the recently discovered phenomenon of noise-induced loss of tracking in feedback systems with PI controllers, anti-windup, and sensor noise.