The stability of implicit self-tuning control has been proved, for the discrete-time linear case, by the use of a Lyapunov func-tion. Latter on the algorithm was extended for a class of bilinear systems. However real world systems are mostly nonlinear systems and it is of interest to extend the proposed algorithm to a more complex class of nonlinear models. In this research a nonlinear class of systems is defined, and then a generalized minimum variance control for the defined nonlinear class is developed. In addition, parameters of real world systems may change in time, and a good performance controller should be able to keep the overall system stability in such a case; to deal with this issue an implicit self-tuning control for the defined class of nonlinear systems is presented, the estimated parameters do not need to converge to their real values. The mathe-matical results show that with this new algorithm the self-tuning controller is able to keep the closed-loop system global stability for the defined class of nonlinear systems, and also the algorithm is a general case of the algorithms proposed in the literature for the bilinear and linear systems cases.Palabras clave: Generalized minimum variance, nonlinear systems, self-tuning control, sliding mode control.ResumenLa estabilidad de los controladores auto-ajustables ha sido demostrada, en el caso lineal discreto, usando una función de Lyapunov. Luego este algoritmo fue extendido a la clase de sistemas bilineales. Sin embargo, en el mundo real los sistemas en su mayoría son del tipo no lineales, por lo que es de gran interés extender el algoritmo propuesto a una clase más compleja de modelos no lineales. En esta investigación se define una clase de sistemas no lineales, y luego a esta clase se le desarrolla un controlador de mínima varianza generalizada. Además, en los sistemas reales los parámetros pueden cambiar en el tiempo, y un buen controlador debe ser capaz de lograr un buen desempeño y mantener la estabilidad global del sistema en lazo cerrado incluso en estos casos. Es por ello que se presenta un controlador auto-ajustable para tratar con las incertidumbres en los parámetros de la clase de sistemas no lineales ya definida, donde los parámetros estimados no necesariamente deben converger a los valores reales. Los resultados matemáticos demuestran que con este nuevo algoritmo el controlador auto-ajustable es capaz de mantener la estabilidad global del sistema en lazo cerrado, y además este algoritmo es un caso general que abarca los algoritmos antes presentados en la literatura para el caso de sistemas bilineales y lineales.Palabras clave: Mínima varianza generalizada, sistemas no lineales, controlador auto-ajustable, control por régimen deslizante.The stability of implicit self-tuning control has been proved, for the discrete-time linear case, by the use of a Lyapunov func-tion. Latter on the algorithm was extended for a class of bilinear systems. However real world systems are mostly nonlinear systems and it is of interest to extend the proposed algorithm to a more complex class of nonlinear models. In this research a nonlinear class of systems is defined, and then a generalized minimum variance control for the defined nonlinear class is developed. In addition, parameters of real world systems may change in time, and a good performance controller should be able to keep the overall system stability in such a case; to deal with this issue an implicit self-tuning control for the defined class of nonlinear systems is presented, the estimated parameters do not need to converge to their real values. The mathe-matical results show that with this new algorithm the self-tuning controller is able to keep the closed-loop system global stability for the defined class of nonlinear systems, and also the algorithm is a general case of the algorithms proposed in the literature for the bilinear and linear systems cases.