In this paper, the separation of binary gas mixture carbon monoxide and hydrogen using single-walled carbon nanotubes (SWNTs) is studied by grand canonical Monte Carlo (GCMC) simulation. All of the particle-particle interactions between hydrogen, carbon monoxide and carbon are modeled with Lennard-Jones potential. Widom test particle method in NVT ensemble is used in determining chemical potentials of the two components of the fluid. it can be concluded from the adsorption isotherms at 293.15 K that the selectivity of carbon monoxide decreases with the increase of the bulk pressure. In the adsorption isostatics at 0.157 MPa, the selectivity has a maximum along with the increase of temperature. However, to given pressure and temperature, the selectivity fluctuates violently with different van der Waals (VDW) distances or diameters of carbon nanotubes. For example, at 0.157 MPa, 293.15 K, diameter of 0.892 nm and VDW distance of 1 nm, with the feed containing hydrogen and carbon monoxide both 50 mol%, carbon monoxide could be enriched up to 98 mol% in the adsorbate, so that the selectivity comes to 60. It can be indicated that carbon nanotube, as a novel material developed in the past 10 years, can make a very effective separation tool of certain gas mixtures.