This dissertation deals with formulations, algorithms and exact methods for solving the well-known Generalized Assignment Problem (GAP) through a price-and-cut approach with the separation of (3, 0.5)-SRC inequalities in order to improve column enumeration feasibility and efficiency. This work is motivated by the perspective of reaching state-of-the-art performance, attaining competitive results which are comparable with the best known solutions found in the literature by Avella (2010) and Michelon (2012). This research was build on exact methods and some heuristics with emphasis on the Dantzig- Wolfe decomposition, the column generation algorithm, the stabilization through weighted Dantzig-Wolfe decomposition proposed byWentges (1997) and finally the column enumeration motivated by the gap minimization reached through the price-and-cut algorithm. The price-and-cut algorithm proposed here resort to column generation (pricing) combined with the separation of (3, 0.5)-SRC cuts in order to increase the generated lower bound, thus minimizing the attained gap. This column generation algorithm follows the work of Savelsbergh (1997); and the separation of (3, 0.5)-SRCs is formulated by Jepsen (2008) and motivated by the branch-cut-and-price algorithm proposed by Poggi and Uchoa (2016) for the CVRP. According to computational experiments, the adopted inequalities are capable of sufficiently reducing the gap, assuring the feasibility of column enumeration for several GAP instances with up to 200 tasks and 20 machines. This method achieved expressive results, compatible with the best known solutions, enumerating all the necessary columns to cover the gap found by the price-and-cut. Therefore, these results motivate future research towards the extension of the method s applicability to larger and more complex instances.