This paper addresses the implementation of a rough controller with proportional–integral action to control a nonlinear system. The system chosen was a level process. Through practical tests, mathematical models (linear and nonlinear) are obtained from the real data of the process. The control system dynamic response in question will be compared with two approaches via discrete linear control and a fuzzy controller. The nonlinear system is a small-scale plant, common in some industrial applications. The rule-based controllers (similar to fuzzy controllers) have linguistic interpretations, are robust with regard to parametric uncertainties of the system, and have a good capacity to map nonlinear characteristics of operation. The fuzzy controllers use procedures of fuzzification and defuzzification of data that require time in the computation of the associated rules. The rough controllers do not require these procedures, requiring less computational time in the corresponding rules, which is an interesting characteristic for real-time applications. This paper presents results of computer simulations and practical tests.