During the past several years, switching control from the viewpoint of both theory and practical implementation has been applied successfully to multivariable systems to accomplish a wide variety of tasks. The motivation for using this type of controller is that, often in practice, no single mathematical model of the system to be controlled is either suitable or available, since the plant is described by a family of plant models. Conventional methods of adaptive controller design generally require that certain types of specific a priori plant information be known and that certain assumptions hold, and thus cannot necessarily be implemented when the plant is described by a family of plant models. In this paper, a new class of simple robust multivariable switching control algorithms is presented to solve the servomechanism problem, which differs from previous works in this area, in that it is not necessary to know the actual family of plant models. The switching controller requires less a priori system information than previously assumed, and only requires that the plant have a satisfactory controller contained in a specified family of controllers. Simulation results obtained when using such a switching controller indicate that a reasonable system transient response generally can be achieved; moreover, experimental application studies using one such class of controllers (with almost no a priori plant information) when applied to a real-time multivariable industrial system additionally are encouraging, in the sense that certain desirable integrity features have been observed to occur.