A naevus is a collection of melanocytes / naevus cells which form a benign skin lesion and appear in different shapes, sizes and colours. They can occur in different parts of the human body and can be more prevalent in areas exposed to sun exposure. Studying naevi is important, as the number of naevi is the strongest melanoma risk predictor. The main aim of this study is to generate realistic looking naevi and to classify them as suspicious and non-suspicious naevi using generative adversarial networks (GANs). This study could be an efficient approach to the early detection of melanoma by identifying suspicious naevi. Two GAN models were explored to implement this research including Deep Convolutional generative adversarial network (DCGAN), Auxiliary Classifier Generative Adversarial Network (ACGAN). We show that an Auxiliary Classifier GAN (ACGAN) achieved high average accuracy, specificity, sensitivity, precision, and AUC with and without augmenting the suspicious naevi images while being able to generate high quality and realistic looking naevi. The ACGAN model also has achieved higher classification outcomes even compared to those incorporating pretraining.