PurposeThe Chinese firms are keenly focused on reducing their environmental footprints as part of the competitive strategy. Within the context of sustainable organizations in China, we test a multilevel framework that examined the impact of environmental-specific servant leadership on the green individual (pro-environmental behavior) and team (project green performance) outcomes within projects. Using social identity theory, we theorize and test the mediating role of green self-identity (individual level) and team green identification (team level) in the relationships between environmental-specific servant leadership, pro-environmental behavior and project green performance.Design/methodology/approachWe used survey questionnaires to collect multi-level and multi-wave data from 42 ongoing project-based sustainable organisations in China. The multilevel team to individual-level hypothesis were analyzed using multilevel-modeling via Mplus, while team level hypotheses were tested using ordinary least squares regression.FindingsThe multilevel regression analysis showed that environmental-specific servant leadership has a trickle-down effect of green self-identity, which subsequently predicts pro-environmental behavior. The ordinary least squares regression results demonstrated that environmental-specific servant leadership predicts project green performance via team green identification. Also, environmental-specific servant leadership has a positive and direct impact on pro-environmental behavior and project green performance.Research limitations/implicationsWe offer community and service dimension of leadership as a determinant of environmental performance at multiple levels. We provide managerial and policy implications to Chinese organizations striving to reposition themselves as eco-friendly organizations both nationally and globally.Originality/valueThe study is among the first to understand the role of environmental-specific servant leadership in predicting individual-level and team-level environment-related mediator and outcomes simultaneously.