Class-imbalanced datasets are commonly observed in hyperspectral image classification and deep learning models often tend to demonstrate a bias toward the dominant class, resulting in very low efficiency for minority classes. Data augmentation has been the most widely used strategy to mitigate class-imbalanced challenge. Therefore, in this paper, the authors have introduced a data augmentation approach utilizing a multi-class generative adversarial network (MCGAN) to deal with this issue. In the proposed approach a generative adversarial network is utilized to specifically generate fake minority class samples and integrating these samples with the existing class samples. Further, a convolutional neural network (CNN) has been employed for classification of hyperspectral image samples. A series of experiments, utilizing the proposed approach, has been conducted with two most popular and publicly available datasets. The experiment results obtained are then compared with four state-of-the-art data augmentation techniques. The comparison results prove the supremacy of the proposed approach in terms of the quality of the generated samples and the classification accuracy over the state-of-the-art techniques.