An overview of the performance of GGA, hybrid and LAP density functionals for transition states in a variety of benchmark systems is given. Examples, such as hydrogen abstraction and intramolecular proton transfer in malonaldehyde, illustrate the difficulties which even the most advanced density functionals meet in describing transition states. Comparisons are also reported for the oligomerization of acetylene Fe(C 2 H 2 ) + 2 --> Fe(C 4 H 4 ) + .Advances in computational techniques and the vast experience of the past decades have allowed an understanding of catalytic reactions at a very detailed level, involving full geometry optimization, characterization of transition states (TS) and reaction intermediates (1-3). The contemporary methods of density functional theory (DFT), especially with the latest progress in the development of efficient exchange-correlation (XC) functionals, appear rather successful in these cases, embodying a large amount of electron correlation at a much lower computational cost (1-3). However, in some cases the reaction path and especially the structure and the energy of the TS may present a stringent test for the existing approximate XC functionals. A typical example is the hydrogen abstraction reaction going through one of the seemingly simplest TS structures, a linear three-atomic complex. The correct reproduction of the energy barrier for this reaction turns out to be a very difficult task not only for approximate DFT methods but also for Hartree-Fock (HF) and most of the post-HF ab initio methods (4,5). In this chapter we shed some light on why this linear three-center TS structure is so difficult to describe.Another big challenge for DFT and post-HF methods is the correct description of inter-molecular and intra-molecular proton transfer (PT) energetics. The energy barrier for the internal proton transfer (IPT) in malonaldehyde is particularly difficult