Trace amounts of carbon monoxide, typically as low as 10 ppm CO, have a deleterious effect on the activation overpotential losses in proton exchange membrane (PEM) fuel cells. This is because CO preferentially adsorbs on the Pt electrocatalyst at the anode at typical PEM fuel cell operating temperatures, thereby preventing the absorption and ionisation of hydrogen. The inability of current preferential oxidation steps to completely remove CO from hydrogen-rich gas streams has stimulated research into CO tolerant anodes. As opposed to other CO oxidation catalysts, metal oxide supported gold catalysts have been shown to be active for the afore mentioned reaction at low temperatures, making it ideal for the 80°C operating temperatures of PEM fuel cells. The objective of this study was to investigate the viability of incorporating titanium dioxide supported gold (Au/TiO2) catalysts inside a PEM fuel cell system to remove CO to levels low enough to prevent poisoning of the Pt-containing anode. Two distinct methods were investigated. In the first method, the incorporation of the said Au/TiO2 catalyst inside the membrane electrode assembly (MEA) of a PEM fuel cell for the selective/preferential oxidation of carbon monoxide to carbon dioxide in hydrogen-rich gas fuels, facilitated by the injection of an air bleed stream, was investigated. It was important for this study to simulate typical fuel cell operating conditions in an external CO oxidation test rig. Factors such as gold loading, oxygen concentration, temperature, pressure, membrane electrode assembly constituents, water formation, and selectivity in hydrogen-rich gas streams, were investigated. The Au/TiO2 catalysts were prepared via deposition-precipitation, a preparation procedure proven to yield nano-sized gold particles, suggested in literature as being crucial for activity on the metal oxide support. The most active catalysts were incorporated into the MEA and its performance tested in a single cell PEM fuel cell. The catalysts proved to yield exceptional activity for all test conditions inside the CO oxidation test rig. However, no significant improvement in CO tolerance was observed when these catalysts were incorporated into the MEA. It was concluded that the thin bilayer configuration resulted in mass transfer and contact time limitations between the catalysts and the simulated reformate gas mixture. Other factors highlighted as possible causes of deactivation included the deleterious effect of the acidic environment in the fuel cell, the formation of liquid water on the catalyst’s surface, and the adverse effect of the organic MEA constituents during the MEA production procedure. The second method investigated was the incorporation of the Au/TiO2 catalyst in an isolated catalyst chamber in the hydrogen feed line to the fuel cell, between the CO contaminated hydrogen gas cylinder and the anode humidifier. Test work in a CO oxidation test rig indicated that with this configuration, the Au/TiO2 catalysts were able to remove CO from concentrations of 2000 ppm to less that 1.3 ppm at a space velocity (SV) of 850 000 ml.gcat -1.h-1 while introducing a 2 per cent air bleed stream. Incorporation of this Au/TiO2 preferential oxidation system into a Johnson Matthey single cell PEM fuel cell test station prevented any measurable CO poisoning when 100 and/or 1000 ppm CO, 2 per cent air in hydrogen was introduced to a 0.39 mg Pt.cm-2 Pt/C anode. These results were superior compared to other state of the art CO tolerance technologies. An economic viability study indicated that the former can be achieved at a cost of gold equal to 0.8 per cent of the USDoE target cost of $45/kW. This concept might allow fuel cells to operate on less pure hydrogen-rich gas, e.g. from H2 that would be stored in a fuel tank/cylinder but that would have some CO contamination and would essentially be dry. The use of less pure H2 should allow a cost incentive to the end user in that less pure H2 can be produced at a significantly lower cost.