The fuzzy supervisor is a rule-based interpolator that allows an improvement in the control performance of nonlinear processes by incorporating different sets of conventional controller gains in multiple operating regions. Based on a proposed static gain ratio, this paper presents systematic methodologies for tuning two fuzzy gain scheduling architectures, namely, in function of the error and its derivative - FGS-f (e, de)-, and in terms of the output and the reference of the process -FGS-f (y, r)-. Behaviors with changes in settling time and oscillations were considered when changing the operation points. The proposed methodologies were successfully tested in a nonlinear simulation model and a continuous flow stirred-tank reactor model. The results show that the FGS-f(e, de) works significantly better than the FGS-f(y, r) and the classic PID controller, in terms of performance indexes based on error accumulation and temporal behavior.