Many reports have focused on breath analysis of breast cancer patients. However, only a few studies have investigated biomarkers through volatile organic components (VOCs) analysis in vitro. The aim of this work is to investigate the volatile organic components fingerprint of human breast cancer cell lines and human normal mammary cells and explore potential VOCs biomarkers for noninvasive diagnosis of breast cancer. MCF-7, MDA-MB-231 and CCD-1095Sk cell lines were employed in this work. Five biological replicates of each cell type were prepared for analysis. Gas chromatography and mass spectrometry (GC-MS) combined with solid-phase microextraction (SPME) was used to detect the VOCs released from the target cell lines. AMDIS-Spectconnect and metabolomic analysis were performed to process the data. According to the results, each kind of cell line shows a unique chromatogram. By applying Principal Component Analysis (PCA), Partial Least Squares Data Analysis (PLS-DA), four components, including 2-ethyl-1-hexanol, 2,4-dimethyl-benzaldehyde, cyclohexanol and p-xylene, were found to be potential biomarkers for discriminating breast cancer cell lines of the HER negative subtype and normal mammary cell line. The combination of the four components was more appropriate for the clinical diagnosis of breast cancer. The study indicated that the VOC profiles of breast cancer cell lines and human normal mammary cell lines were quite different and the investigated biomarkers may hold promise for non-invasive diagnosis of breast cancer.