Abstract: Dynamic model parameter identification of accelerometer plays an important role in improving the accuracy of dynamic testing and analysis for vibration and shock. To overcome the influence of the fence effect (FE) on the accuracy of the parameters obtained with dynamic model parameter identification method for accelerometer frequency domain, a dynamic model parameter identification method of accelerometer based on discrete spectrum correction and least square (DSC-LS) was presented. Firstly, the point coordinates at zero frequency of accelerometer were estimated by using <i<H</i<<sub<1</sub< estimator, and the resonant point coordinates were estimated with higher accuracy by combining FFT+FT discrete spectrum correction and LS. Then, according to the feature points coordinates, the dynamic model parameters of accelerometer were calculated. The experimental results show that, the presented method can eliminate the influence of FE efficiently on the accuracy of accelerometer's dynamic model parameters identification and improve the anti-interference capacity for low frequency noise with narrow-band, and improve the identification accuracy of the dynamic model parameters of accelerometer.