Density functional theory has become a valuable tool to study surface catalysis. However, due to the scarcity of clean and reliable experimental data on surfaces, the theoretical methods employed to explore heterogeneous catalytic mechanisms are usually less well validated than those for gas-phase reactions. We argue herein that gas-phase reactions and the corresponding surface reactions are related through the Born–Haber cycle and computational catalysis on surfaces will be less meaningful if gas-phase behavior cannot first be suitably determined. In this contribution, we have constructed a set of gas-phase reactions relevant to the Fischer–Tropsch synthesis as a case study. With this set, we have tested the validity of the widely used PBE and B3LYP functionals and found that neither of them are capable of describing all kinds of gas-phase reactions properly, such that some surface reactions may be biased falsely against the others. Significantly, XYG3, which is a double-hybrid functional that includes Hartree–Fock-like exchange and many-body perturbation correlation effects, presents a significant improvement for all of the gas-phase reactions, holding promise for further development for surface catalysis.