Despite the well-established health benefits of the Mediterranean diet, there are signs that Mediterranean populations are deviating from this traditional pattern. We aimed to evaluate longitudinal changes in adherence to the Mediterranean diet, its determinants and health effects in a representative sample of the adult Greek population. This was a secondary analysis of the ATTICA epidemiological cohort study conducted in 2001/2002 and 2011/2012. The study sample consisted of 3042 men and women free of cardiovascular diseases living in Attica, Greece; of them, 2583 were followed-up for 10 years. Participants were evaluated in terms of sociodemographic, lifestyle and clinical parameters at baseline, and incidence of cardiometabolic diseases was recorded at follow-up. Dietary habits were assessed both at baseline and 10 years through a validated food frequency questionnaire and adherence to the Mediterranean diet was evaluated through the MedDietScore, based on which four trajectories were identified, i.e., low–low, low–high, high–low and high–high. During the study period, 45.6% of participants moved away from the Mediterranean diet (high–low), 9.0% moved closer (low–high), while 18.7% sustained a high adherence (high–high). Participants in the high–high trajectory were younger, mostly women, more physically active, had a higher socioeconomic status, and a more favorable body composition and cardiometabolic profile at baseline, and exhibited lower 10-year incidence rates of hyperlipidemia, hypertension, diabetes mellitus and cardiovascular disease compared to other trajectories (all p-values < 0.050). Adherence to the Mediterranean diet is declining among Greek adults. Staying close to the Mediterranean diet is associated with significant health benefits and should be a major target of public health strategies.