The immune defense and the repair function of the pulp tissue serve as the biological foundation of pulpotomy. The precise evaluation of the pulp inflammation extent and determining its reversibility are essential for the success of pulpotomy. The objective of this study was to classify the molecular-level of dental pulp cell physiology and inflammatory state based on the biochemical changes obtained by single-cell Raman spectroscopy. Firstly, we differentiated the growth of HDPCs (human dental pulp cells) under physiological states by employing Raman spectroscopy with multivariate statistical analysis. Raman spectroscopy reflected the biochemical changes at different growth phases, including the lag phase, log phase, and stationary phase. Secondly, we evaluated the optimal concentration and duration of Porphyromonas gingivalis lipopolysaccharide (P.g.LPS) stimulation to establish a six-level inflammation classification model of HDPCs. Thirdly, we performed label-free characterization of biological component changes in cells of different inflammation grades by Raman spectroscopy. As a result, the differences of peaks in the region 600-1800 cm-1 demonstrated the biochemical molecular alterations in the different inflammation grades of HDPCs. As inflammation progresses in steps, protein peaks increased first and then decreased, while lipid and nucleic acid peaks gradually decreased compared to unstimulated cells. However, when the inflammatory stimulation reached grade V, the changes in the biological properties were characterized by a recovery in protein and lipid content, and a decrease in nucleic acid content. We then established the diagnostic model using the Raman spectra of HDPCs in physiological and inflammatory states, which had a prediction accuracy of 100 % and 97.4 %, respectively. Finally, we determined the reversibility threshold of HDPCs at different grades of inflammation. We observed that the inflammation of grade I and II cells had potential reversibility and could be attempted to be retained. In conclusion, Raman spectroscopy combined with multivariate statistical analysis has potential possibility to effectively distinguish the degree of inflammation in the dental pulp, thus providing new tools and perspectives on pulpotomy in clinical practice.