The distinguishment between the tumor tissue and healthy tissue is a key step during tumor surgery. Raman spectroscopy is a promising tool to achieve this function. However, the blood around the tissue poses challenges in determining the boundary of the tumor tissue. In this paper, the blood effect on the Raman spectral of breast tumor tissue is investigated. The Raman spectra of the tumor tissue and its adjacent healthy tissue were collected. Then, principal component analysis (PCA) is adopted to reduce data dimension, and support vector machine (SVM) is used to classify tumors and healthy tissue. The blood effect index (BEI) is proposed to evaluate the effect of blood on the Raman spectra of breast tumor tissue and healthy tissue. The effect of blood on the Raman spectra is analyzed. It demonstrated that the existence of blood could affect the intensity of the Raman peaks. The affected peak was selected and processed with the recursive least-squares (RLS) algorithm in order to reduce the effect of blood on the spectra.