Abstract: Single-cell metabolomics has emerged as a powerful tool to decipher cellular heterogeneity and uncover hidden biological processes at the individual cell level. However, achieving high-throughput and sensitive detection of metabolites from single cells remains a significant technical challenge. In this study, an integrated microfluidic chip-mass spectrometry platform (μCyESI-MS) was developed that enables label-free, real-time, and high-throughput metabolic profiling of single cells. The μCyESI-MS system utilizes a spiral microfluidic chip fabricated from polydimethylsiloxane (PDMS), featuring two inlets and one outlet for cell introduction, sheath fluid injection, and analyte collection, respectively. The 10-loop spiral channel efficiently aligns and focuses cells into a single-cell stream. Each cell is lysed on-chip by the sheath fluid, and the released intracellular metabolites are directly transferred into an electrospray ionization mass spectrometer (ESI-MS) for immediate analysis. This approach achieves a throughput of approximately 30 cells per minute, with minimal signal overlap and excellent detection sensitivity. This platform was applied to profile the metabolic fingerprints of three widely used human cancer cell lines, including HeLa, HepG2, and MCF-7. By matching accurate mass values with metabolite databases, a variety of metabolites were detected and identified, including tyrosine, carnitine, phosphocholine, glycerophosphocholine, acylcarnitine, glutathione, monoglyceride MG16, long-chain acylcarnitine, and lysophosphatidylcholine (LysoPC 16:0). Lipids, especially phosphatidylcholines (PCs) and sphingomyelins (SMs), are important components of the cell membrane and exhibit high ionization efficiency under positive ion mode. As a result, they are more readily detected by mass spectrometry, showing strong and specific signal responses. Dimensionality reduction using t-SNE revealed that the three cancer cell types could not be clearly separated in the small-molecule metabolite range (m/z 300-650), whereas the lipid-rich region (m/z 650-900) exhibited significant discriminatory power for cell-type classification. Ten representative lipid molecules with significant differences were selected, including PCs plasmalogen phosphatidylcholines (PC Ps), and SMs. Their relative abundances in HeLa, HepG2, and MCF-7 cells were compared, and their potential as biomarkers for cancer cell typing was evaluated. This study presents a robust and scalable platform for high-throughput single-cell metabolomics analysis. The μCyESI-MS system integrates precise microfluidic cell handling with sensitive MS detection, offering new opportunities for exploring metabolic heterogeneity, discovering biomarkers, and improving our understanding of disease mechanisms at the single-cell level.