This paper presents an investigation on the performance of shunt active power filter (Shunt APF) systems based on five-level neural point clamped (NPC) inverter in steady and transient conditions adopting a fuzzy control scheme. The shunt APF is operated as a current source connected in parallel with the non-linear loads to inject compensates current harmonics into the AC source. The multi-level inverters have gained considerable interests in last year’s. Five-level inverter is recommended in medium and high voltage applications; their advantages are low harmonic distortions, low switching losses, low electromagnetic interference, low voltage stress of power semiconductors and low acoustic noise. Secondly fuzzy logic has been widely used in various industrial applications. To benefit from these advantages a shunt active power filter based on five-level (NPC) inverter using fuzzy control scheme is presented in this work. To identify the reference currents, synchronous detection method (SMD) is adopted. Proportional integral voltage controller is used to maintain the dc voltage across capacitor constant and reduce inverter losses. The simulation is performed using MATLAB-Simulink software and SIMPOWERSYSTEM toolbox. The obtained results in steady and transient states illustrate the performances and the effectiveness of the proposed shunt APF system.