To achieve the goal of low-cost MEMS gyros for the precise self-localization of mobile robots, this paper presents a simple, yet effective method to minimize drifts on the heading angle by combining measurements from a gyro with measurements from wheel encoders (odometry). The main idea of the proposed approach is to estimate the accuracy of both sensors as a function of the actual maneuver being carried out, and then the output of both sensors are fused by the complementary filter taking into account the maneuvering conditions. The proposed method is applied to a mobile robot and the experimental data demonstrate the effectiveness of this approach.