The near collapse out-of-plane (OOP) deformation criterion is critical for the seismic assessment of unreinforced masonry (URM) walls by an implicit nonlinear time-history analysis (NLTHA). Therefore, different methods, i.e., experimental tests, nonlinear kinematic analysis (NLKA) and NLTHA, are referred to in this study. Results by NLKA show that the near collapse OOP deformation is largely influenced by geometry, boundary conditions, overburden load and the initial drift, etc. Meanwhile, there are many differences between one-way and two-way OOP bending performances of URM walls. Comparison and discussion are specifically performed for these two scenarios in the paper. Results show that all above mentioned factors have some influences on the OOP deformation capacity for both scenarios. Furthermore, some dynamic experimental tests are analyzed by NLKA and NLTHA in the paper. It shows that NLKA gives relatively reasonable estimation, while NLTHA by DIANA software has a slight overestimation of near collapse OOP deformation capacity. Therefore, it is difficult to set a simple criterion for all the cases. However, some conservative conclusions on the OOP deformation criteria, drawn hereafter based on the features of Groningen buildings, can be used for the application of NLTHA.