There have been many studies of hot surface ignition of premixed gases. However, using the experimentalresults obtained up to this time, it is difficult to understand the basic mechanism of ignition by heated surfaces because these experiments were made under the normal gravity condition, suffering from the effect of natural convection. In the present study, experiments were made on the ignition of mixtures by heated nickel, tungsten,and platinum wires under normal gravity and microgravity conditions. With a methane-oxygen mixture ignited by a heated tungsten wire, a strong gravity effect on the ignition delay is observed. On the contrary, there is almost no gravity effect observed with a methane-air mixture. Results of calculations show that ignition experiments under the microgravity condition are simulated successfully by this numerical model. It is shown from the numerical results that the ignition point in the methane-oxygen case is apart from the hot-wire surface, while, in the methane-air case, this point exists near the hot-wire surface. Therefore, natural convection has little effect on the ignition delay of a methane-air mixture. Experimental results of hydrogen-air mixtures ignited by heated platinum wires show that chemicalspecies are supplied to the platinum wire surface by natural convection, and therefore, the surface reaction is promoted under the normal gravity condition. Experimental and numerical results show that the surface reaction is the most energetic when the equivalence ratio of the mixture is 0.3, which conflicts with the result obtained by Coward and Guest that the rate of catalytic reaction attains a maximum at stoichiometric composition. This contradiction may be due partially to the gravity effect in their experiments.