It is predicted that crops like watermelon will be decreased in coming years and price of this product in the market will increase. This highlights the necessity of measures to choose high-quality watermelons by the final consumer. Also according to the concept of virtual water, sorted and more desirable watermelons can be exported at higher prices in order to create more benefit. In general, the aim of this study is to provide a measure that is based on Charleston Gray watermelon morphological characteristics and evaluation of the classification ratio in unripe, ripped and overripe classes, by data mining algorithms. The results of the sensory evaluation showed that experts (human) were able to classify 52% of the samples correctly. The correct classification algorithm K Nearest Neighbor was significantly higher than the classification of LVQ Neural Networks and Discriminant Analysis but classification results of different distance metrics of this algorithm showed no significant difference using them. The highest correct classification with the amount of 67.3 percent belonged to Support Vector Machine algorithm with Gaussian kernel function. Although at first glance it may seem that, this amount is far from ideal but it should be noted that this amount is 15% higher than the classification made by humans. Incidentally, this classification was done based on morphological characteristics of samples which measuring them does not require sophisticated tools and methods.