Abstract: Rapid evaporative ionization mass spectrometry (REIMS) is a novel mass spectrometry technology that does not require any sample pretreatment. Mass spectrum could be easily acquired by a hand-held intelligent knife coupling to REIMS to obtain the lipidomics profile. In this study, REIMS, principal component analysis (PCA), and orthogonal projection to latent structure (OPLS) were used to real-time assessing the lipidomics profiles of pomfret (Pampus argenteus) during air-frying and explore the effect of different air-frying temperatures on the lipid composition of pomfret muscle tissue. The parameters of the REIMS system were optimized by single factor experiment, including the output power 20 W, cutting power 1 mm/s, and auxiliary solvent flow rate 100 μL/min, under which a stable and reliable mass spectrogram was obtained. The PCA and OPLS analysis were carried out, and there were 10 important characteristic ions were obtained (VIP>1). Among the featured ions, PE (O-16∶1/22∶6) was detected only in raw material with proportion 4.61%, whereas not detected in the fried sample. In addition, PE (22∶5/22∶6) as the highest unsaturation of the 10 phospholipid ions, the relative content of which decreased with the increase of the frying temperature, also seen in the PA (18∶1/22∶6) and PI (18∶0/22∶6), indicating that the saturation of phospholipids grew with increasing frying temperature. The method validation showed that the signal to noise ratio of selected ions was from 49.58 to 205.30 and the relative standard deviation of intra-day and inter-day accuracy was 3.10%-7.28%. REIMS is sensitive and accurate, so that it has great potential to be a simple, efficient, and accurate novel technical mean in the process of lipidomics research.