A fuzzy logic method is developed for gain scheduling of PID controllers that improves upon existing fuzzy-PID schemes. The method uses a single fuzzy input variable related to the derivative of the PID manipulated variable, eliminating redundant rules. Online tuning improves PID performance while retaining original PID parameters. The fuzzy and PID manipulated variables are related through a differential equation, allowing online replacement and tuning with two parameters. The method is demonstrated on a temperature control process, improving PID performance to the level of model predictive control with only a few tuning tests.