Serum cytokines are promising biomarkers of cancer staging and outcome prediction, including response to treatment. Serum samples were collected from 200 breast carcinoma patients prior to chemotherapy treatment. Luminex liquid protein chip technology was used to analyze 25 cytokines in serum. Linear regression was used to analyze the relationship of cytokine levels and tumor size. The independent sample T-test and Chi-square test methods were used to analyze the difference of cytokine levels between two groups. IL-12p40, sIL-2R, MMP-2 levels showed linear correlation with tumor size. Eotaxin, IL-10, IL-12p70, IL-7, IL-1ra, IP-10, MCP-1beta, MP-2 and MIP-1beta levels showed significant difference between different lymph node groups, but only Eotaxin, IP-10 and MCP-1 levels had an inverse correlation with the number of positive nodes. Fractalkine, G-CSF, MIP-1alpha, MIP-1beta levels showed significant differences between different ER+ groups. Eotaxin, Fractalkine, IL-6, IL-7, IL-10, MCP-1 and VEGF levels had significant differences between different HER-2 groups. Our study resulted in the identification of a serum cytokine profile with the potential to be clinically applicable to predict disease outcome and in monitoring of efficacy of treatment.