This letter proposes a power system dynamic state estimator (DSE) by integrating measurement correlations. The correlations between voltage phasors and calculated real/ reactive power as well as the correlations between real and reactive power at the generator terminal buses are analyzed. The unscented transformation approach is used to estimate the measurement error covariance matrix that models the self-and cross-correlations. By integrating these correlations into the unscented Kalman filter (UKF)-based DSE, we show that it achieves higher statistical efficiency than the conventional UKF in the presence of both small and large measurement noise.