This paper analyzes the rice production technology in Korea in the context of multi-output framework. The estimation uses a set of national farm-level panel data within the primal approach. The particular focus of the study is to analyze the existence of heterogeneous technologies for different farm sizes, and the possibility of output substitution. Unlike many previous studies, we employ a farm-specific fixed effects model, which accounts for the effects of individual specific characteristics or constraints on production. Our results suggest that considerable differences in technology exist across various farm sizes. Interestingly, input substitution and output substitution possibilities are more limited for smaller farms. However, in general, output substitution possibilities are very limited across all farm sizes.