By investigating the transmission of uncertainty and information in linear time-invariant multivariable control systems disturbed by stationary stochastic processes, performance limits of disturbance rejection are studied within the framework of information theory. Two measures of information and uncertainty, entropy rate and mutual information rate, are employed as performance functions of linear regulation systems and linear tracking systems, respectively. For the linear regulation problem, the entropy rate of the system output is computed by using the Bode integral formula, then the performance limit of disturbance rejection is formulated in terms of unstable poles of the open-loop transfer function, and the “conservation law of variety” is revised. For the linear tracking problem, a relation between the \(H_\infty\) entropy of the system’s closed-loop transfer function and mutual information rate of the pair of disturbance and output is deduced by using a frequency calculation method for the mutual information rate, and an upper bound of disturbance rejection performance in Gaussian systems is derived based on this relation.