Here we describe the successful application of our recently developed computational enzyme design methodology [1] to construct new protein catalysts for reactions for which no naturally occurring enzyme exists. Starting from a 3-dimensional description of the minimal active site (consisting of the transition state of the desired reaction surrounded by side chain functional groups optimally oriented for catalysis), the Rosetta-Match and RosettaDesign algorithms screen a large library of existing protein structures and identify and design pockets to accommodate the minimal active site. The enzymes which are predicted to catalyze the reaction of interest are then expressed in bacteria, purified and assayed in vitro to confirm activity. Thus, the method is general enough to be potentially applied to any chemical reaction. We successfully applied this technique to three reactions for which no natural occurring enzyme exist: the Kemp elimination reaction [2], a retro-Aldol reaction [3], and a Diels-Alder reaction [4]. The presented data will focus on the Diels-Alderase, which represents the first computationally designed protein catalyst for a bimolecular reaction. [1] A. Zanghellini et al., Protein Science, 2006, 15, 2785 [2] D. Ršothlisberger et al., Nature, 2008, 453(7192), 190 [3] L. Jiang et al., Science, 2008, 319(5868), 1387 [4] A. Zanghellini et al., to be published