Purpose: We aimed to develop noninvasive and early detection breast cancer biomarkers panel that may serve as assistant diagnostic method.Methods: 61 biomarkers were detected in sea of 101 healthy controls, 46 benign breast diseases and 77 breast cancer patients in the training group. A metropolis algorithm with Monte Carlo simulation was used for choosing the model. 444 individuals were used for validation. Serum from 245 female cancer patients including 5 kinds of cancers were also collected to evaluate cancer selectivity. Results: Panel consisting of Apolipoprotein A І (ApoA І ), ApopB, C-reactive protein (CRP) and interleukin (IL)-8 had the highest value for discriminating between breast cancer and healthy control. The sensitivity (SN) was 98.70% for all-stage, 100.00% for early-stage and 97.92% for advanced-stage with 90% specificity (SP). In the validation group, the sensitivities were 96.43%, 100.00% and 94.21% at 90% SP. This panel identified 14.29% cervical cancer, 0% lung cancer, 20.29% pancreatic cancer, 25.00% gastric cancer, and 17.50% colorectal cancer as non-breast cancer. Panel consisting of Pepsinogen (PG) І /II, CRP, Superoxide dismutase, Tumor necrosis factor α had the highest value for discriminating between breast cancer and benign breast diseases. The SN was 88.31% for all-stage, 72.41% for early-stage and 97.92% for advanced-stage with 90% SP. In the validation group, the sensitivities were 81.25%, 69.77% and 88.41% at 90% SP.Conclusions: The biomarker panels showed an improved performance when compared to CA153. It may serve as assistant tools for breast cancer screening and early detection to improve the clinical outcome.