Many studies have shown that improvements in the numerical prediction about rolled sheet forming is done through the laws of behaviour increasingly complex, particularly by the combination of isotropic and kinematics hardening (mixed hardening) to take on the Bauschinger effect. But there are several types of test experimental characterization Bauschinger effect: simple shear in one direction and then reverse, tension and compression, bending unbending test… In the case of steel, this paper will answer these two questions: Are these experimental tests are equivalent in terms of Bauschinger effect? What is the impact of these different tests on the identification of parameters of the law of behaviour?