In a study of more than 100 NSCLC cases we previously showed that the expression of BARD1 isoforms is correlated with poor patient survival. BARD1 is a tumor suppressor acting with BRCA1 as ubiquitin ligase. BARD1 has also functions in mitosis and poly (ADP)-ribose signaling for DNA repair. In cancer cells BARD1 isoforms are generated by alternative splicing. SNP affecting splicing and cancer predisposition were identified in neuroblastoma. The alternatively spliced isoforms lack tumor suppressor functions and act as oncogenes. As the domain composition of cancer-associated isoforms predicts altered tertiary structures, we investigated whether BARD1 isoforms act as cancer antigens. ELISA assays were performed to detect antibodies generated against BARD1 isoforms in the serum of lung cancer patients. We used BARD1 protein fragments and short peptides for capturing autoimmune antibodies. Using fitted Lasso logistic regression methods, we developed an algorithm for the prediction of lung cancer based on a blood test for detection of BARD1 antibodies. Modeling values from 200 samples, shows a distinction of lung cancer and healthy controls with high sensitivity and specificity (AUC=0.961; Figure 1). Splitting the samples randomly and repeatedly into training sets and validation sets, confirmed an average AUC=0.964 for the training sets and AUC=0.861 for the validation sets. ROC curves for early and late stage lung cancers showed no difference in their AUCs. The BARD1 lung cancer test is highly specific and does not cross-react with other cancers. Lung cancer has a very long latent phase and is often discovered at an advanced and untreatable stage. Currently the detection by low dose CT scan is relatively expensive and not very specific. Therefore a blood test, such as the BARD1 test could i) help to detect cancers earlier, in particular by screening of risk groups, and ii) become a diagnostic aid in combination with CT scan.