To ensure that a space nuclear power system will operate safely and respond in a predictable and desired manner, the system’s controller design must account for changes in the system parameters over its lifetime. A model reference adaptive controller is applied to enable the actual space nuclear power system to follow a predictable and desired response of a reference model system, despite changes in the actual system’s operating parameters. Model reference adaptive control is well developed for linear systems and has been applied to simple, single-input, single-output (and the output’s derivative) systems. Model reference adaptive control is applied to a single-input, multiple-output nonlinear system but also shows the development for a multiple-input, multiple-output linear system. An algorithm is developed for linear systems to determine the constant gains in the model reference adaptive control algorithm and a method is developed that allows selective weighting of a desired state variable. Examples are presented to show that a model reference adaptive controller can ensure the load-following response of a nonlinear space nuclear power system and that the reference model can be complex enough to embody the physics of the plant. The results of the example cases show that a model reference adaptive controller can cause a selected nonlinear plant state variable to track the transient trajectory of the corresponding state variable of the reference model with local stability.