The effects of the sampling and the interpolation of strong motion data on the corrected accelerograms and integrated displacements are investigated in both frequency and time domains, It is found that the sampling of a record can be consideied as a low-pass filter which eliminates the high frequency contents in the record, but the interpolation of data does play a role of high frequency noise source which introduces high frequency noise into digital records. In is also founa that an unequally spaced sampling model will give a more accurate information than an equally spaced sampling model at the same sampling rate. But the latter will result in a wider range of flat frequency response than the former, than the former would. In comparison with the results given by the linear interpolation method, the parabolic one can generally provide with much better results, especially when a lower sampling rate is used.