Most of preliminary evaluation methods to accomplish the seismic vulnerability of bridges consider as an evaluation parameter the substructure irregularity. This parameter is estimated by means of the difference in the length of piers or through the pier typology, but both evaluated by a subjective form. In this paper the evaluation of the influence of substructure irregularity is presented. First, a simple and regular bridge, with three piers of equal length and four spans, is considered as an original structure. Starting from this system, different irregular models were elaborated reducing or incrementing the length of the central or one of the external piers, in percentages of 25%, 50% and 75%. As a seismic action, a database of more than 50 earthquakes, registered in the Mexican Pacific Coast, the most seismic hazardous zone of México, were considered. The selected registers have magnitudes greater than six and important peak ground accelerations. By means of elastic analyses, the variation between regular and irregular maximum responses in displacements and internal forces are determined. Of the obtained responses, statistical values as mean and standard deviation are evaluated. With these values, percentages of difference in the response of irregular bridges were estimated; these percentages could be included as fragility weights of the irregularity degree in more rigorous preliminary methods to evaluate the seismic vulnerability of bridges.