The nitrogen reduction reaction (NRR) using new and efficient electrocatalysts is a promising alternative to the traditional Haber-Bosch process. Nevertheless, it remains a challenge to design efficient catalysts with improved catalytic performance. Herein, various O-functional MXenes were investigated as NRR catalysts by a combination of density functional theory calculations and least absolute shrinkage and selection operator (LASSO) regression. Nb 3 C 2 O X has been regarded as a promising catalyst for the NRR because of its stability, activity, and selectivity. The potential-determining step is *NH 2 hydrogenation to *NH 3 with a limiting potential of –0.45 V. Furthermore, via LASSO regression, the descriptors and equations fitting the relationship between the properties of O-functional MXenes and NRR activity have been proposed. This work not only provides a rational design strategy for catalysts but also provides machine learning data for further investigation. A three-step screening method (including stability, activity, and selectivity screening) was applied to develop NRR catalysts from O-functional MXenes, and LASSO regression was applied to describe the original NRR performance.