The game of poker has received increasing attention in computer science, as it constitutes a model for decision making in an incomplete information environment. In the thesis a variety of evolutionary and coevolutionary techniques are investigated so as to generate computer poker players close to the Nash equilibrium. Specifically, we are concerned with the well-known poker variant Two-player (Heads Up) No Limit Texas Hold'em. We begin with the evolution of players for the first phase (pre-flop) of the game using an exact probability model for the win chances of a player. The player is represented by an evolved action table, which holds specific actions to be played in specific situations. Using pre-flop play as a test bed a promising coevolutionary environment is developed, which then is extended to the complete game (player) by a bootstrap approach, where the evolution of a player for a specific game phase is based on the best evolved players for previous phases.