To overcome the linearizated errors from the attitude estimation algorithm based on Extended Kalman Filter ( EKF), a new attitude estimation algorithm based on gravity/magnetic assisted Euler angle Unscented Kalman Filter ( UKF)algorithm was proposed to improve the attitude measuring accuracy of a low cost Micro-electro-mechanic System(MEMS). The gravity and magnetic data were used to inhibit the rapid divergence of attitude error for the MEMS. The Euler angles were taken as UKF states and the quaternion was used to calculate the attitude in time update, so that the algorithm avoids the quaternion standardized problem and solves the low attitude accuracy of Euler angles. Without linearization errors of the UKF, it has better stability and attitude estimation accuracy. By taking measured MEMS Inertial Measurement Unit (IMU )data for experiments, the results show that the measuring accuracy of pitch and roll angles by proposed algorithm has improved nearly 20 percent respectively and the heading accuracy improved by 12.1 percent as compared with that of the attitude estimation algorithm based on the EKF. It concludes that the proposed method has better precision. However, the convergence time of UKF has increased due to the insufficient estimation for state variances.