Background Determining disease activity in systemic lupus erythematosus (SLE) patients is challenging and limited by the lack of reliable biomarkers. Analysis of the altered distribution of circulating B cells has shown promise for assessment disease activity. Yet studies are limited by variable classifications of B subsets. Therefore, we studied peripheral B cells using high-dimensional tools to identify a valid biomarker of disease activity.Methods We studied B cells in two separate cohorts of patients included in the Swiss SLE Cohort Study. In discovery cohort A, we analyzed cryopreserved PBMCs from 30 SLE, and 30 age-, sex- and ethnicity matched healthy controls (HC) by mass cytometry. In validation cohort B, fresh blood from 63 SLE, 14 Sjögren syndrome (pSS), 14 Sarcoidosis (Sarc), and 39 age-matched HC were analyzed by flow cytometry.Results In cohort A, using unsupervised clustering analysis, we identified 7 metaclusters within B cells. Two metaclusters were increased in SLE and exhibited a phenotype of atypical memory B cells (aMBC): CD21-, CD27-, CD11c+ and CXCR5-. Based on cohort A results, we confirmed in cohort B the increase in CD21-CD27- aMBC in SLE, compared to healthy and disease controls. In both cohorts, aMBC were associated with the severity of clinical manifestations. Compared to classical biomarkers, aMBC showed a significant correlation with clinical signs of disease activity. Conclusion aMBC were expanded in SLE, and the increase correlated with clinical disease activity. According to our data, aMBC represents a robust and easily accessible biomarker to assess disease activity in patients with SLE.