The elastic contact problem, as implemented in some commercial software such as ANSYS, depends on the user choice of some parameters such as normal contact stiffness, penetration limit and contact algorithms. This work investigates the artificial neural networks (ANN) potential to predict the value of some parameters, avoiding the trial-and-error procedure to determine these values. Contact problems based on simple problems are used to train the neural network, so it can predict the normal contact stiffness for more complex problems. Some contact examples are evaluated, including the small end connecting rod contact problem, of great importance in automobile industry. (C) 2015 Elsevier B.V. All rights reserved.97 4353Thyssen Krupp Metalurgica Campo Limpo Ltda (Campo Limpo - Sao Paulo - Brazil)