This research project attempts to improve the understanding of oil bank formation in Fontainebleau sandstone cores using a surfactant polymer (SP) EOR method. While oil bank formation is generally successful in longer length cores, it is elusive in short cores (<10cm). Core flooding experiments were performed in a specially designed setup allowing tests to be done on cores with varying length and rock properties. The cores were saturated with a model oil, dodecane, and the injected surfactant polymer slug was optimised using phase behaviour testing. The first objective was to see whether the chosen SP could efficiently mobilise the residual oil left after water flooding and create an oil bank. This was successful in the 17 and 30cm cores as expected, the tests were also successful for the shorter (7cm) cores. A small amount of tests were also done under a CT-scanner. Using iododecane as the dopant, the displacement of oil was visualised at regular intervals during the flooding stages. The images confirm oil bank formation and also gave insight into the rock properties. Although the cores were chosen for its homogeneous properties, many of the cores showed fingering and two layer displacement which would indicate otherwise. Recovery factors averaged around 80\% for the low permeability cores and up to 90\% for the higher permeabilities. Production data was the main source of output data and subjected to various methods of analysis with a focus on effective relative permeabilities. Using the analytical JBN approach which is normally applied on water-flood data, this relatively simple method was tested to see whether it could produce accurate relative permeability curves which are essential to EOR analysis. Current standard practice involves matching pressure and production data which is both complex and time consuming. The resulting relative permeabilities show that under identical displacement processes the surfactant polymer mobilises oil more freely in cores with higher pore volume. The results from the JBN method were compared to results using Shell’s in-house numerical simulator (MoReS). For most core lengths the results from the simulator confirmed the relative permeability curves obtained using the JBN method. In the experiments with large discrepancies further studies should be done to determine whether this was caused by an error in the method or by the heterogeneities in the cores as seen on the CT scan. Using this approach could offer an efficient and robust alternative to the current technique of establishing RelPerm curves.