This paper presents power flow optimization using UPFC based on neuro-fuzzy method for multi-machine power system stability. The controllers of power flow must be able to control both active and reactive power flow without reducing the stability of the power system. The recent power flow technology is the use of a Unified Power Flow Controller (UPFC). The common transmission line parameters, i.e. impedance, voltage, and phase angle, can be controlled simultaneously by the UPFC. UPFC could provide the different functions when the inserted voltages have different in magnitudes and phase angles. In the literature, some traditional techniques used in UPFC are still firmly with logic. These techniques have difficulties to multi-machine systems which is very dynamic. Therefore, the method based on neuro-fuzzy is implemented to the adaptive UPFC for adaptively controlling the power flow in dynamic multi-machine system in this study. The three phase fault is done to the multi-machine system. The adaptive neuro-fuzzy UPFC controller show that the overshoot occurred during the fault is low.