We evaluated the role of the presence of non-alcoholic fatty liver disease (NAFLD) at baseline in the transition from metabolically healthy to metabolically unhealthy obesity (MHO to MUO) ten years later. A prospective cohort study (ATTICA study, Greece) was performed between 2002 and 2012 studying a sample from the greater metropolitan Athens area. In total, 1514 (49·8%) men and 1528 (50.2%) women (aged >18 years old) free-of-CVD were included. Healthy metabolic status was defined as absence of all NCEP ATP III (2005) metabolic syndrome components. NAFLD was defined according to validated liver steatosis indices. Follow-up CVD assessment (2011–2012) was achieved in n = 2020 participants (n = 317 cases). NAFLD prevalence among MHO participants ranged from 29% to 39% according to the specific NAFLD score used. MHO participants who developed metabolically unhealthy status had about two times higher odds to have NAFLD at baseline compared with their metabolically healthy normal weight counterparts whereas stable MHO was not associated significantly with NAFLD. Moreover, MHO status accompanied by NAFLD was associated with increased CVD risk (Hazard Ratio = 2.90 95% Confidence Interval (1.35, 5.40)) in comparison to their non-NAFLD MHO counterparts. Further analysis revealed that in the obese, NAFLD indices and not simply visceral adiposity increased significantly the ability of metabolic status (using standard definition) to predict long-term CVD incidence. Considering NAFLD, even when assessed using validated indices only, in the clinical assessment of apparently healthy obese individuals predicts who is to develop MUO and contributes independently and more accurately to defining future cardiometabolic risk. •MHO individuals had equal chances to have NAFLD compared with MUO subjects.•MHO participants with NAFLD had 2 times higher chance to become MUO in long-term.•This transition subsequently increased their cardiometabolic risk.•NAFLD inclusion in metabolic status definition of obese could be of added value.