In a cohort of atrial fibrillation patients, to classify patients in different groups according to their propensity of being correctly anticoagulated (weekly probability of presenting an INR between 2 and 3) over their first year of treatment with Vitamin K Antagonists using Group Based Trajectory Models (GBTM). GBTM, a type of latent class analysis, can suppose an alternative method for summarizing INR control by incorporating information on its dynamic nature, providing a classification of patients into different trajectories over time, described through friendly graphics. Finally, we aimed to assess clinical outcomes in each of the trajectories. Real-world, population-based cohort of AF patients initiating with VKA in the region of Valencia from 2010 to 2015 and on treatment over the whole year after initiation. We used GBTM to identify weekly trajectories, modeled with quadratic polynomial functions of time. We calculated the Time in Therapeutic Range associated to every trajectory ultimately identified and we assessed their association with clinical effectiveness and safety outcomes (mortality, stroke, hameorragic stroke, bleeding). We included 7,971 patients in the cohort that fulfilled the inclusion criteria. Mean number of INR determinations over the first year of treatment was 13,9. Combining formal statistical model fit indices (AIC, BIC, Lo-Mendell-Rubin likelihood ratio test) and usefulness criteria (substantive interpretation, minimum size and discriminative properties—entropy), we identified four differential trajectories of INR control: optimal (10.3% of patients, TTR: 83.7), improving (27.8% of patients, TTR: 61.3), worsening (28,5%; TTR: 69%) and uncrontrolled (33,5%; TTR: 41.3). In unadjusted analysis, mortality was significantly higher in the group of uncontrolled patients. Differences in other outcomes did not achieve statistical significance. Only 10% of patients were optimally controlled over the first year of VKA treatment. Mortality was higher in uncontrolled patients. GBTM are useful to target patient subgroups that are more likely to benefit from improvement interventions.