This research aims to classifying (clustering) the results of catching fish per month for the period to 2015 to 2017 by using the k-means algorithm and determine the most superior fish.Therefore the existence of this research by using the algorithm of K-means clustering can assist to provide the information about superior species of fish or the most abundant species and fish that appear less in the sea waters of Ternate, in order to facilitate the fishermen in preparing for arrest next fish. Data taken by the PPN Ternate is the monthly fishing catch data for the 2015 to 2017 period. Cluster evaluation methods use the DBI method (Davies Bouldin Index) to find out how good or good clusters are used in the research, the authors have calculated several clusters, namely 2 cluster, 3 cluster, 4 cluster, and 5 cluster to determine the comparison, the result is the DBI value in 2 cluster which is closest to value of 0. The DBI value obtained in 2 clusters in 2015 = 0.395, year 2016 = 0.276, and in 2017 = 0.54. The results of the research used 18 fish data to produce 2 clusters, which there were 16 data entered cluster one (C1) and 2 data that goes into the cluster two (C2.