To the measurement error caused by sensor due to rotor vibration in attitude estimation for quad-rotor aircraft, the data measured by accelerometer and gyroscope are fused through a Kalman filter method, to improve the accuracy of attitude estimation, reduce the noise caused by carrier vibration and solve the gyro drift problem.First, the error types of inertial devices were analyzed, and then a three-DOF (degree of freedom) system model was established, simulation analysis was also made through Matlab.Finally, the feasibility of this method was verified by using an experimental bench with propellers.Experimental results showed that the attitude estimation error was within 2° and was not accumulated over time, basically meeting the requirements of attitude estimation for quad-rotor aircraft, such as real-time performance, low cost and high precision.This method could be successfully applied to the attitude control for quad-rotor aircraft, which can suppress the noise interference, and create a good condition for stable autonomous flight.