This paper presents a laboratory experiment to investigate how urban costs might determine the internal structure of urban areas (monocentric or polycentric) by influencing location choices of firms and households. The experimental design is in part guided by a theoretical model that shows how the trade-off between communication costs faced by firms and commuting costs borne by workers determine the degree of cities' polycentricism by influencing the distribution of the workplaces within it. In the laboratory experiment, groups of 16 subjects participated in a two-step auction game, where firms and workers first negotiated to find a job contract and second where workers bid for land in order to find a home. The game is repeated for four rounds and different experimental treatments are implemented, each defining a given scenario for communication costs and commuting costs. The chosen benchmark is a polycentric city treatment, where no communication cost exists for firms, giving them no incentive to locate in the CBD. In two other treatments, the communication cost is positive. In the Monocentric City treatment, commuting cost for workers is low, giving a clear incentive for firms to locate in the CBD, as workers do not suffer from potential high commuting costs, giving an outcome where all firms should locate in the CBD. In the Hierarchical City treatment, the commuting cost is very high for workers, giving workers a strong incentive to live close to their workplace and accepting lower wages as a result. Experimental results are in line with theoretical predictions: firms tend to locate in the CBD under the monocentric treatment, whereas more distant locations are accepted leading to a hierarchical outcome as workers propose lower wages for not working in the CBD so as to escape high commuting costs.