This paper presents a comprehensive and integrated databank of the Iranian strong ground motions that occurred from 1973 to 2018. The databank consists of 7196 three-component acceleration records from 3180 earthquakes and 1157 stations in Iran. In this paper, the characteristics of the Iranian strong ground motion data are presented in terms of event, station, and recording distributions. The events are characterized by magnitude in the range 2.4–7.7. Shear wave velocity has been measured and reported at 603 strong motion stations of the databank. In this study, three different empirical techniques are applied to classify the stations. A new method is proposed for site classification based on the correlation coefficient between the horizontal-to-vertical (H/V) response spectral ratios of the ground motion records recorded by each station. It is noticeable that the raw accelerograms have been uniformly processed in the entire databank using the filtering and wavelet de-noising methods to remove high- and low-frequency noise. Moreover, by comparison between the Fourier Amplitude Spectrum (FAS) of the noises detected in all acceleration and velocity time series by the filtering and the wavelet de-noising methods, it was determined that the mean and mode of FAS of the noises detected by the wavelet de-noising method in most of the frequencies is higher than mean and mode of FAS of the noises detected by the filtering method.