AbstractCarbon fiber reinforced thermoplastic (CFRTP) composites, known for their excellent mechanical properties and environmental sustainability, are highly sought‐after in aerospace, energy, and many other fields. Grinding could be used to realize the high‐accuracy and low‐damage machining of CFRTP components. However, due to the temperature sensitivity and water absorption of CFRTP, dry and flood lubrication grinding could cause thermal damage and hygrothermal aging, respectively. In addition, these methods could lead to dust and excessive cutting fluids harmful to the environment. Addressing these bottlenecks, this study is the first to propose using oil‐on‐water minimum quantity lubrication (OoW‐MQL) to grind CFRTP. Experiments including various lubricating conditions and grinding parameters were performed to investigate the grinding performance. Experimental results showed that OoW‐MQL with the oil–water proportion of 1:1 outperformed the other lubricating conditions, achieving the best surface quality with the lowest grinding forces and temperatures. The optimal oil–water proportion remained almost unchanged under different grinding parameters. Furthermore, importance analysis using the extreme gradient boosting (XGBoost) algorithm demonstrated that oil–water proportion was the most critical factor affecting surface roughness. This study proves the feasibility of OoW‐MQL and provides technical guidance for high‐quality grinding of CFRTP.Highlights Oil‐on‐water minimum quantity lubrication (OoW‐MQL) was proposed to grind carbon fiber reinforced thermoplastic (CFRTP) composites. The experimental results of grinding forces, grinding temperatures, surface roughness and surface morphologies were analyzed, which proved that OoW‐MQL with oil–water proportion (1,1) had outstanding lubricating and cooling capability. The effect of grinding parameters on the optimal oil–water proportion was proved to be negligible through comparative tests under grinding parameters of high force and high temperature. Quantitative analysis using the extreme gradient boosting (XGBoost) algorithm identified the oil–water proportion as the most critical factor influencing surface roughness.