With the rapid development of economy and industrialization, global warming is becoming the most serious sensitive global climate issues, which causes the rising of sea level and many other negative effects. The cause of global warming is the emission of greenhouse gases and carbon dioxide is the main component of greenhouse gases. The control of CO2 emssion is beneficial to addressing gobal climate change and environmental degradation. Therefore, it’s important to develop a rapid detection of CO2 for accurate control. There are amounts of methods to detect CO2 at present, including titration, electrochemical method, gas chromatography, infrared absorption spectroscopy and so on, however, t they still have the deficiency for online monitoring in industrial field. laser induced breakdown spectroscopy (LIBS), which is developing rapidly in recent few decades, is a detecting technology with characteristics of time-saving and synchronous measuring of multicomponent. What’s more, there is no need for sample pretreating. To develop the online monitoring technique of CO2 emission in the industrial field, LIBS was employed to measure CO2 in this study. The mass flow controller was used to adjust the flow of high purity CO2 and N2 to obtain mixed gas with different CO2 concentrations. The mixed gas was firstly mixed in an air mixing chamber for thorough mixing and then sent to the sample cell for LIBS measurement. The evolution of C atomic spectral line and CN molecular band with different delay times were being studied, which demonstrated parts of CO2 react with air ambient to form CN molecular during the plasma generation, the CN molecular band should be taken into consideration for quantitative analysis, and the parameters were optimized for synchronous measurement of C line and CN band: 800 ns was the optimal delay time. During the plasma generation, many factors in the plasma may interact with others, the analysis index had close relationship wih serval measuring parameters. With the consideration of the effect of C, CN and the self-absorption in high concentration, multivariate calibration method was employed to establish calibration models of CO2. The results showed that the correlation coefficients R2 and the slope were 0.978 and 0.981, respectively. Compared with calibrated with single factor, the multivariate method improved the reliability of the model. What’s more, the feasibility of the application of LIBS to measure CO2 rapidly was proved.