The diagnosis of acute leukemia is based on a series of essays. Cytogenetics and fluorescence in situ hybridization are designed to identify specific genes and chromosomal changes involved in the development of leukemia. Immunophenotypic analysis by flow cytometry and molecular biology is also present. It can be used to check the diagnosis. The invasiveness and timeliness of these tests prompted people to look for new methods and techniques, represented by Raman spectroscopy, to reduce the time interval between hypotheses and diagnostic conclusions, and to help patients achieve good prognosis. Raman spectroscopy can be used to non-label identify spectral information of blood cells and biochemical elements present in the blood, such as amino acids, proteins, lipids, nucleic acids, and carotenoids, and the technique is performed using cell lines, blood smears, and serum samples. The diagnosis of different types of hematological diseases becomes more and more important and has broad application prospects. This experiment investigated the Raman spectral characteristics of human acute T cell leukemia cell line (Jurkat), human acute myeloid leukemia cell line (KG-1α), human acute promyelocytic leukemia cell line (NB4) and human acute monocytic leukemia cell line (THP-1), build a novel Raman label-free method to distinguish four kinds of acute leukemia cells, and provide the basis for clinical experimental research. Raman spectra were obtained using Horiba Xplora Raman spectrometer, and Raman spectra of 25~30 cells from four kinds of acute leukemia cell line were recorded. Spectral analysis methods, including principal component analysis, partial least squares discrimination analysis and cluster analysis were used to build the discrimination model and classify the spectra of leukemia cells. There were differences among Raman spectra of acute leukemia cells, which were distinguished primarily by peaks in 769, 826, 844, 957, 1 048, 1 141, 1 255, 1 313, 1 415, 1 539, 1 575 cm^-1, which reflects the active metabolism of Jurkat cells, the nucleic acid-related energy metabolism of KG-1α cells, and the strong cellular respiration of NB4 cells. The identification models built by PLS-DA can accurately distinguish these Raman spectra of different acute leukemia cell lines. The model has good fitting stability, predictive ability and application prospect.