Background: Earlier detection of cancer could lead to improved outcomes. Non-invasive biomarkers are integral for early detection, and protein and DNA-based assays have been previously described for this purpose. Evaluation of highly expressed circulating RNA would improve upon existing methods by allowing for quantitation, however RNA expression in plasma is low. Circular RNA (circRNA) is formed by alternative backsplicing and provide resistance to exoribonclease-mediated degradation. Therefore, circRNA is more stable in plasma and could potentially be more easily detected non-invasively. Here, we report results from a novel assay that identifies and quantitates expression of circRNA in cancer plasma. Methods: Cancer-specific circRNA targets were identified by analyzing differential expression and abundance based on capture RNA sequencing in tissues. CircRNA-specific probes were designed and validated by stable expression with RNAse R incubation. Expression of selected circRNA isoforms were measured using a quantitative PCR-based assay. For breast cancer specimens, plasma was collected from a cohort of 36 patients and 12 normal female controls without known cancer diagnosis. Expression was determined by normalizing signal to two averaged universally expressed circRNAs. Differential expression for each target was determined for cancer vs. normal and stratified based on ER and HER2 positivity as well as stage. Results: Normalized expression of seven proposed breast cancer-specific circRNA targets was determined for all patients in the cohort. All targets demonstrated resistance to RNAse R-mediated degradation in breast cancer cell lines. In the breast cancer cohort, CircTRPS1 expression was higher in cancer plasma than normal, trending toward significance. Expression of circTRPS1 was significantly more common in ER-negative than ER-positive breast cancer (p = 0.02). CircPDHX expression was elevated in normal plasma compared to cancer and relatively higher in ER-positive than ER-negative cancer (p = 0.05). Median expression in plasma was elevated in cancer specimens compared to normal for CircVAV3, circDNAH14, and circCDYL. CircERBB2 expression was also relatively higher in HER2-positive cancer when compared to other targets. Higher expression significantly positively correlated with metastatic disease for circERBB2 (p = 0.01) and circCDYL (p = 0.005) and negatively correlated with metastasis for circWDR78 (p = 0.02). Conclusions: CircRNA are promising biomarkers for early non-invasive detection of cancer. We identified seven circRNA biomarkers in breast cancer plasma which correlated with clinical characteristics. Three circRNAs were significantly prognostic for metastatic breast cancer. Identification of additional cancer-specific targets and multiplexing markers is underway. This assay is generalizable to other cancers, and similar studies in prostate and bladder cancer are in progress.