In this paper, we propose a linear quadratic regulator (LQR) for a kind of three-phase two-level voltage source inverter on the basis of grid voltage modulated-direct power control (GVM-DPC) principle. The proposed method has a similar control structure to the conventional GVM-DPC scheme, but the algorithm parameters tuning problem therein could be solved by LQR. Considering the inverter system often converges to a nonzero equilibrium point, an auxiliary variable corresponding to the tracking error is necessary. Unfortunately, when utilizing the LQR technology to solve the optimal tracking problem directly, the dynamic response and steady-state performance might be poor in some situations. By developing a novel auxiliary variable, the drawbacks aforementioned can be avoided and the optimal feedback gain matrix only needs to be calculated once during the control process offline. Then, a Lyapunov stability analysis is used to prove the exponential stability of the system. Finally, simulation and experimental results validate the performance of the proposed method. Note to Practitioners-With the high integration of renewable energy in modern power systems, inverters have become the core device for the transition among electricity, wind, and solar energy. This trend makes the research of power converters regarding control problems increasingly important. Although existing studies can solve some basic problems such as controlling the output current or power of inverters, most of them are based on phase-locked loops, and controller parameters are not easy to tune. This motivates us to develop a user-friendly, low computational burden, easily-to-tune parameters control algorithm, which is convenient to implement on embedded devices to control the output power of inverters. Inspired by this, a direct power control method for inverters using linear quadratic regulator is proposed in this study, and a phase-locked loop-less method is developed by utilizing the theory of grid voltage modulation. Furthermore, this method establishes a new auxiliary variable to ensure global exponential stability of the linear quadratic regulator, resulting in satisfactory transient performance. The above advantages have been further validated through experiments and compared with existing algorithms.