Abstract: Strong motion records of ground have been used as the ground motion input for the elastic-plastic time series analysis of engineering structures. It requires necessary data processing before used, as they are obtained by instrumental observations. Many strong motion records of ground have been accumulated in China, but there is a lack of standardized procedure of data processing, causing some inconveniences to users and devaluating their scientific applications. A complete procedure of data processing for the strong motion record of ground is proposed, which will be particularly used in engineering practice. Three principles for selecting the cut-off frequency of high-pass filtering are suggested, including that the low-frequency Fourier amplitude spectrum must conform to the w² source model, that the signal-to-noise ratio must be greater than 3, and that the physical resolution keeps consistent. The effects of causal and acausal filtering on the recorded time series are investigated. The results show that these two approaches have their own characteristics, causal filtering causes phase shift and acausal filtering causes interference vibration before signal arrives. The effects of causal and acausal filtering on the pseudo acceleration response spectrum are investigated using data collected in Lushan earthquake in 2013 together with some data recorded in 2017. It shows that the phase shift caused by causal filtering can cause bias in the calculation of PSA in the effective period range. Thusly, it is recommended to use acausal filtering for strong motion records used in engineering practice. It could help to guide the use of strong motion records of ground in an appropriate way.