Abstract: With the methods of single-specimen and multiple-specimen, the subsequent yield surfaces of the copper underwent tension, torsion and combined tension torsion pre-deformation respectively are studied by crystal plasticity simulation comparing with the real test employing thin-walled tubular specimens. Not the variances of measured subsequent yield surface but also the phenomena of subsequent yield surface appearing concave are discussed under considering different conditions, including different pre-deformation, the number of probing point, the test sequence of the probing points, and the specified offset strain, etc. On this basis, the rationality and validity of the two methods for subsequent yield testing are compared. The simulations probing yield surface are conducted using the FE model of thin-walled tubular specimen, in which the crystal lattice orientation for each element is arranged randomly, associated with a modified crystal plasticity constitutive model that is able to reflecting the Bauschinger effect for material under reversed loading, so that the deformation characteristic of polycrystal can be exhibited. The loading procedure of modeling test is designed as same as the real test. The investigated results show that:(1) the proposed simulation method can reproduce the real test procedure, the simulated subsequent yield phenomena are found fairly consistent with that observed in actual experimental measurements, which confirmed the rationality and validity of the method; (2) both the simulated and real tests show that the subsequent yield surface measured by using the thin-walled tube under combination of tension-torsion load may be concave, and the result tested by the single specimen method is more obvious in yield surface concave no matter for simulated or real test; (3) if the test material is fairly consistent in quality, the multiple-specimen method should be more reasonable than the single-specimen method for subsequent yield surface test.