The label printing problem (LPP) arises in the printing industry, where we have a set of labels that require printing. Each one has different demands and must be printed using a single template. The number of labels we can allocate in a template is a fixed one, and the number of templates varies in a small range of different sizes. The goal of the LPP is to design templates to meet the demands for the labels while minimizing the percentage of material wasted. In each grid, the need for each printed cover must be met to facilitate label management. To deal with LPP, we used two known metaheuristics: a threshold accepting and a tabu search. These heuristics, with the aid of a Hill climbing double neighborhood, improved the results significantly.