In the past, milling operations have been mainly considered from the economic and technological perspectives, while the environmental consideration has been becoming highly imperative nowadays. In this study, a systemic optimization approach is presented to identify the Pareto-optimal values of some key process parameters for low-carbon milling operation. The approach consists of the following stages. Firstly, regression models are established to characterize the relationship between milling parameters and several important performance indicators, i.e. material removal rate, carbon emission, and surface roughness. Then, a multi-objective optimization model is further constructed for identifying the optimal process parameters, and a hybrid Non-dominated Sorting Genetic Algorithm-II algorithm is proposed to obtain the Pareto frontier of the non-dominated solutions. Based on the Taguchi design method, dry milling experiments on aluminum are performed to verify the proposed regression and optimization models. The experimental results show that a higher spindle speed and feed rate are more advantageous for achieving the performance indicators, and the depth of cut is the most critical process parameter because the increase of the depth of cut results in the decrease of the specific carbon emission but the increase of the material removal rate and surface roughness. Finally, based on the regression models and the optimization approach, an online platform is developed to obtain in-process information of energy consumption and carbon emission for real-time decision making, and a simulation case is conducted in three different scenarios to verify the proposed approach.