We have performed a miscible CO 2 flood study using fractured and unfractured carbonate cores and 31° API west Texas oil to evaluate oil recovery under three main injection modes: SWAG, WAG, and CGI. For each injection mode, three cases were considered: unfractured, one horizontal fracture and two fractures (one horizontal and one vertical). Increasing the number of fractures will investigate different shape factors such as the elongated slab and sugar cube models. Then, a commercial simulator was used to match the experimental results and model the fractures. The experimental results may be summarized as follows. First, for the unfractured case, the oil recoveries for SWAG, WAG, and CGI are 98, 92, and 75 % of OOIP, respectively. Second, for the elongated slab case (one horizontal fracture), the oil recoveries for SWAG, WAG and CGI decreased to 72, 70, and 27 %, respectively. Third, for sugar cube model (two fractures), the oil recoveries for SWAG, WAG and CGI are 79, 79, and 55 %, respectively. The simulation shows that the improvement in recoveries during SWAG and WAG over CGI is a result of a better conformance provided by the injected water which hindered CO 2 mobility and decreased its cycling through the fracture. Also, the sugar cube model has shown better results than the elongated slab because of the presence of a vertical fracture in the middle which helped CO 2 to diffuse more in the matrix and contact oil. The results suggest that injecting CGI in fractured cases is not recommended and injecting water (wetting phase) along with CO 2 is essential. Finally, among all injection modes, SWAG has resulted in the most uniform displacement profile with the lowest residual oil saturation after the flood. This injection mode combines the benefits of water to transfer into the matrix with CO 2 excellent displacement.