To enlarge the dynamic measurement range and improve the performance by reducing the influence of the main noise including thermo-mechanical noise, 1/f mechanical and tunneling noise, Johnson noise, and shot noise, tunneling sensors usually are operated in a closed-loop mode. Under the assumption that these noises are approximately Gaussian and the tunneling current/gap exponential relationship can be linearized by small-signal linearization, the Linear Quadratic Gaussian (LQG) optimal controller was designed for our micromachined tunneling gyroscope (MTG) to maintain a constant tunneling gap. Simulated results indicated that the distance from the proof mass to the tunneling electrode was effectively regulated to its nominal value of 1 nm thanks to Kalman state estimator associated with the LQG controller. And the closed-control system using LQG control methodology showed an increase of more than 24 dB in signal-noise ratio compared to uncontrolled gyroscope.