Taro is a tropical tuberous vegetable that produces cormels of high nutritional and energetic value. In recent years, its cultivation has expanded, creating the need for more research. However, the experimental plot size interferes directly with the precision of the experiment.Literature describes several methods for establishing the optimum experimental plot size. The objective of the current research was to estimate the optimal experimental plot size for taro culture, utilizing the Modified Maximum Curvature Method. The field experiment was conducted in the vegetable garden of the Departamentode Fitotecnia of Universidade Federal de Viçosa, from September 2011 to June 2012, with Japanese taro. Twenty-two rows with 20 plants each were planted, at 1.0×0.3 m spacing, considering the 20 central rows with 18 plants each as useful area, totaling 360 plants. Each plant corresponded to a basic unit (BU). Once the possible clusters were made, 23 different plot sizes were formed. At harvest, the fresh mass characteristics of large, medium, and marketable cormels were evaluated. The optimum plot sizes were: 8.77 BU for marketable cormels; 14.75 BU for large cormels, and 13.37 BU for medium cormels. Among the plot sizes estimated in this study, the aim was to improve precision of experiments performed with taro, plots formed by 15 BU that corresponded to a 4.5 m2 area are recommended.