Prospective identification of patients with chronic lymphocytic leukemia (CLL) destined to progress would greatly facilitate their clinical management. Recently, whole-genome DNA methylation analyses identified three clinicobiologic CLL subgroups with an epigenetic signature related to different normal B-cell counterparts. Here, we developed a clinically applicable method to identify these subgroups and to study their clinical relevance. Using a support vector machine approach, we built a prediction model using five epigenetic biomarkers that was able to classify CLL patients accurately into the three subgroups, namely naive B-cell-like, intermediate and memory B-cell-like CLL. DNA methylation was quantified by highly reproducible bisulfite pyrosequencing assays in two independent CLL series. In the initial series (n=211), the three subgroups showed differential levels of IGHV (immunoglobulin heavy-chain locus) mutation (P<0.001) and VH usage (P<0.03), as well as different clinical features and outcome in terms of time to first treatment (TTT) and overall survival (P<0.001). A multivariate Cox model showed that epigenetic classification was the strongest predictor of TTT (P<0.001) along with Binet stage (P<0.001). These findings were corroborated in a validation series (n=97). In this study, we developed a simple and robust method using epigenetic biomarkers to categorize CLLs into three subgroups with different clinicobiologic features and outcome.