This paper proposes a new RWO-Sampling (Random Walk Over-Sampling) based on graphs for imbalanced datasets. In this method, two schemes based on under-sampling and over-sampling methods are introduced to keep the proximity information robust to noises and outliers. After constructing the first graph on minority class, RWO-Sampling will be implemented on selected samples, and the rest will remain unchanged. The second graph is constructed for the majority class, and the samples in a low-density area (outliers) are removed. Finally, in the proposed method, samples of the majority class in a high-density area are selected, and the rest are eliminated. Furthermore, utilizing RWO-sampling, the boundary of minority class is increased though the outliers are not raised. This method is tested, and the number of evaluation measures is compared to previous methods on nine continuous attribute datasets with different over-sampling rates and one data set for the diagnosis of COVID-19 disease. The experimental results indicated the high efficiency and flexibility of the proposed method for the classification of imbalanced data