The main purpose of this study is to apply non-parametric method of data envelopment analysis (DEA) for optimization of energy inputs and reduction of greenhouse gas (GHG) emission for rice production in Guilan province of Iran. For these aims, 120 rice producers were selected for data collection in Astaneh Ashrafiyeh city of Guilan province. In this paper, seven energy inputs and rice yield as output were considered for DEA method. The technical, pure technical and scale efficiency were determined based on Charnes-Cooper-Rhodes (CCR) and Banker-Charnes-Cooper (BCC) models. The average of technical, pure technical and scale efficiency were calculated as 0.79, 0.98 and 0.81, respectively. The results revealed 35 (29.17% of total units) and 72 (60% of total units) rice producers were efficient in technical and pure technical efficiency, respectively. The total energy saving was estimated about 19.80%. Also, the highest share of contribution to the total savings energy was calculated as 43.41% for diesel fuel. The energy use efficiency was improved about 25% by converting present farms to target units. Furthermore, the GHG emission of each input was investigated for present and optimum units. The results indicated that the total GHG emission of present and optimum farms was calculated as 1847.26 and 1483.52 kgCO_2eq. ha⁻¹, respectively. Moreover, the effect of energy optimization in reduction of GHG emission was found to be as 363.74 kgCO_2eq. ha⁻¹. With respect to DEA method, the diesel fuel consumption of efficient units had the highest effect to helping for GHG reduction.