For many years durability tests engineers have worked in the sense of improving the tests that, at first, were performed using public roads with high time consumption and low reproducibility. Proving grounds were specially designed to reproduce the most important efforts to the body and chassis systems, but time problem was still there.Time and cost reduction allied to the needs of quality, reliability and reproducibility improvement led the engineers to develop methods and equipments to reproduce the durability tests in the lab. In this way the road simulators appear as a powerful tool able to perform durability tests with high reliability, self-controlled and with very low time compared to the road tests. At this scenery bench tests were also created to components and systems mainly used to anticipate problems before a whole vehicle test.A greater number of bench tests are performed using constant amplitude sinusoidal signals based on a statistical study of efforts on a component or system covering the related variables. However, at the real world, loads come into the vehicle through the suspension not following any sequence and are present in a random form varying from tension to compression with infinity of amplitudes, frequency and cycle average values. Bench tests improvement pass through to put in consideration the inherent random phenomenon. Besides, bench tests must be damage equivalent and be able to reproduce the failure observed at the road.The main purpose of the work was developing a durability bench test to the body region where the exhaust vehicle system is attached. This test should be correlated in damage and failure with the road, to be used as a reliable alternative to evaluate durability behavior of this body region. At end, results from road and bench test were compared.