Experimental and numerical techniques have been developed to evaluate performance of CO2 huff-n-puff processes for unlocking resources from the Bakken tight oil formations. Experimentally, a series of core-flooding tests have been conducted by using core samples collected from the Bakken tight formation with a permeability range of 0.27-0.83 mD. The performance of four recovery schemes (i.e., waterflooding, immiscible CO2 huff-n-puff, near-miscible CO2 huff-n-puff, and miscible CO2 huff-n-puff processes) has been comparatively evaluated. Theoretically, numerical simulation is performed to match the experimental measurements and then extended for evaluating field-scale CO2 huff-n-puff performance in the Bakken formation. Once the history matching on either the experimental measurements or the field geological model is completed, the tuned numerical model is then employed to optimize the injection pressure, soaking time, and production pressure, respectively. It is experimentally found that the waterflooding process leads to a higher oil recovery factor of 51.5% in comparison with the immiscible CO2 huff-n-puff process, while both the near-miscible and miscible CO2 huff-n-puff processes result in a higher recovery efficiency with the final recovery factors of 63.0% and 61.0% of OOIP, respectively. As for the history matched reservoir geological model, either a higher injection pressure or a lower production pressure usually results in better recovery performance but lower than the corresponding experimental measurements, while recovery factor can be optimized if soaking time of 15 days is applied in this study. (C) 2016 Elsevier Ltd. All rights reserved.