Abstract. One of the tasks of the Indonesian Citrus and Subtropical Research Institute is research on crossing between citrus varieties to produce saplings with the best quality products through observation of the fruit produced. Because the amount of fruit production studied is very large, it requires a fast and accurate observation process, one of which is the clustering method of data mining. Observations were made using a clustering process or grouping Density Based Spatial Clustering Application with Noise (DBSCAN) on fruit characteristics that indicate quality. DBSCAN works by grouping data based on density, so that it is expected to find several data groups that are close to each other which shows the tendency of the quality of the observed fruit data as well as labeling outlays for data that are too far from the crowd. The results of the grouping will be analyzed to find out the number and characteristics of the groups formed where the results of the grouping are assessed using the Silhouette Coefficient method to determine the best parameter values. The results obtained in this study are obtained three group results which will be divided into medium quality, good, and not so good. The quality of grouping using the Silhouette Coefficient value of 0.69.