An eight-layer discontinuous sperm isolation medium (PureSperm gradient) was evaluated in separation of human spermatozoa according to sex chromosomes by fluorescence in-situ hybridization (FISH). This study was carried out on spare samples from normozoospermic and oligozoospermic patients referred to the Royan Institute for infertility treatment. Semen analysis was assessed according to World Health Organization criteria. X- and Y-bearing spermatozoa were simultaneously identified in the neat semen (control) and sperm isolation medium fractions from the same samples using FISH and chromosome specific DNA probes. At least 1000 spermatozoa were scored for each sample. The proportions of X- and Y-bearing spermatozoa were determined by presence of red or green fluorescent signals. Before separation, there was no significant difference in the percentage of spermatozoa with the specific signals of X and Y chromosomes. After separation, in both normal and oligozoospermic patients, the percentage of X-bearing spermatozoa in the bottom layer slightly exceeded that in the top layer (P = 0.001). In both the normal and oligozoospermic groups, the difference between the frequencies of Y-bearing spermatozoa in the top and bottom layers was significant (P = 0.001). It seemed that eight-step discontinuous gradient was not a reliable method for the separation X- and Y-bearing spermatozoa for clinical purposes.