Concentrated loads on steel beams are frequently found in structural engineering practice. In situations where the location of the load is fixed, traverse web stiffeners can be used to provide an adequate increased resistance, but for economic reasons should be avoided whenever possible. For moving loads, the knowledge of the unstiffened web resistance becomes imperative. Many theories were developed for a better understanding of the problem, however, a 20% error is still present in the current design formulas. A more accurate design formula for this structural problem is very difficult to be obtained, due to the influence of several interdependent parameters. On the other hand, creating new experimental results is very expensive and time consuming. As an alternative, numerical methods were used to generate new data but they still present significant differences. The neural networks were inspired in brain structure, to present human characteristics such as: learning from experience; and generalization of new data from a current set of standards. These characteristics demonstrate the possibility of using the neural networks to investigate complex problems which could not be solved by conventional programs, such as the forecasting of the ultimate strength of steel beams subjected to concentrated loads. The main objective of this work is to investigate the structural behavior of concentrated loads, by means of a parametrical analysis. This analysis can surely help to identify the influence of the several interdependent parameters. Thus, the possibility of using neural networks to generate new data is investigated. A literature survey of the problem is presented, as well as a critical evaluation of the experimental results to be used as the training data set of the neural networks. In order to evaluate or train the neural networks, the use of finite element simulations will be analyzed. These numerical simulations will be performed using a finite element method interface system which was developed to integrate the preprocessing, analysis and post-processing programs. The use of numerical results present in the literature is also analyzed and discussed. In order to validate the neural networks model, the results are compared with existing design formulas enabling the parametrical analysis to be executed. The neural networks model can also be used as a new tool to forecast the ultimate load of steel beams subjected to concentrated loads.