The finite element method (FEM) is commonly used is solving engineering problems including fatigue crack growth. The S-version FEM is an extended version of the FEM to predict fatigue crack growth. This method was developed into open-source software written in C/C++ language and built based on the UNIX operating system. The objective of this study is to perform a prediction for fatigue crack growth using the S-version FEM. This S-version FEM was structured using the global-local overlay technique that consisted of two separate meshes for global and local. Therefore, the computation process was solely focused on the local area instead of whole geometries. This undoubtedly reduced computation time and dependency on conventional FEM software. The geometry of the problem was modelled in FAST (pre-post) program and referred to as a global mesh, and the mesh was automatically generated in FAST. The location of the crack refers to the local mesh and at this stage smaller elements and finer mesh were made at the crack area. Local mesh was overlaid with global mesh. Then the global-local overlay approach was used to compute the engineering problem. The stress intensity factor (SIF) was calculated using the virtual crack closure-integral method (VCCM). The prediction of the fatigue crack growth results are shown in this study as well as the behaviour of the SIF corresponding to the crack growth. Analytical solutions were compared with the Sversion FEM to validate the results. A very good agreement was presented between the S-version FEM and analytical approach for the normalised SIF. In addition, regression analysis was performed to support the evidence. From the analysis, very small root mean square errors (RMSE) of 0.12 with high correlation coefficient (R 2 ) of 99.74% were shown to confirm the findings. From this study, it can be concluded that the S-version FEM is suitable to be used to predict fatigue crack growth for semi-elliptical surface cracks. The application of the S-version FEM benefits the user because using the overlay approach shortens the computation process.