Abstract: Background and Objective: Rice is the major cereal crop for the people of North Buton and Southeast Asia, as well. To determine the diversity of various local upland rice genotypes based on agronomic traits and yield potential, a field experiment was conducted in villages of the Kambowa district, North Buton Regency, Indonesia. Materials and Methods: Twenty-two local upland rice genotypes were evaluated for their diversity in regard to agronomic traits and yield potential. The experimental plots were laid out in a randomized complete block design (RCBD) with 3 replicates. Statistical analysis of variance (two-way ANOVA) to test the significance and to measure the least significant differences between means was carried out. Results: Based on plant height, 12 cultivars were classified as tall plants and 10 cultivars were classified as short plants. The tallest upland rice plants were Wakawondu and Wangkaluku, while the Wajini cultivar was the shortest one. Among the 22 local upland rice cultivars, 7 genotypes were recorded as having a potential yield higher than 3.0 t ha–1. The grain yield/plant exhibited positive significant correlations with filled grain number (r = 0.910), 1000-grain weight (r = 0.906), grain number (r = 0.890), tiller number (r = 0.753), productive tiller number (r = 0.747) and panicle length (r = 0.635) at a 0.01 significance level. However, plant height and leaf area had a negative correlation with the yield/plant (r = -0.140 and r = -0.092, respectively). Furthermore, yield was observed to be positively associated with panicle bearing tillers and number of filled grains/panicle and these characteristics were noticed to exert strong direct effects on grain yield/plant. Conclusion: It was concluded that filled grain number, 1000-grain weight, grain number/panicle, productive tiller number and panicle length can be considered good criteria for the selection of rice cultivars suitable for crop yield improvement programs.