The ubiquitous presence of the Fe line complex in the X-ray spectra of galaxy clusters offers the possibility of measuring their redshift without resorting to spectroscopic follow-up observations. In this paper we assess the accuracy with which the redshift of galaxy clusters can be recovered from an X-ray spectral analysis of Chandra archival data. This study indicates a strategy to build large surveys of clusters whose identification and redshift measurement are both based on X-ray data alone. We apply a blind search for K--shell and L--shell Fe line complex in X-ray cluster spectra using Chandra archival observations of galaxy clusters. The Fe line in the ICM spectra can be detected by simply analyzing the C-statistics variation $Delta C_{stat}$ as a function of the redshift parameter. We repeat the measurement under different conditions, and compare the X-ray derived redshift $z_X$ with the one obtained by means of optical spectroscopy $z_o$. We explore how a number of priors on metallicity and luminosity can be effectively used to reduce catastrophic errors. The $Delta C_{stat}$ provides the most efficient means for discarding wrong redshift measures and to estimate the actual error on $z_X$. We identify a simple and efficient procedure for optimally measuring the redshifts from the X-ray spectral analysis of clusters of galaxies. When this procedure is applied to mock catalogs extracted from high sensitivity, wide-area cluster surveys, such as those proposed with Wide Field X-ray Telescope (WFXT) mission, it is possible to obtain a complete samples of X-ray clusters with reliable redshift measurements, thus avoiding time-consuming optical spectroscopic observations. This methodology will make it possible to trace cosmic growth by studying the evolution of the cluster mass function directly using X-ray data.