Objective: To investigate diabetes outcomes by long-term trajectories of patients' care settings among diabetes patients with regular follow-up. Research design and methods: This longitudinal population-based cohort study used data from the National Health Insurance Research Database in Taiwan. The trajectories of diabetes care-seeking behaviours among newly diagnosed diabetes patients with regular follow-up were identified using a group-based trajectory model from 2000 to 2004. Severe diabetes complications were assessed for the period 2005–2010. Cox-proportional hazard method with a cumulative incidence function was applied. Results: Among the diabetes patients with regular follow-up during the first 5 years since diagnosis, 44.5% and 40.6% were persistently cared for in generalized care settings and specialized settings, respectively. Only 14.9% of them had shifted between different levels of care setting during the first 5 years. No significant difference in adverse outcomes was observed between patients who were persistently treated in generalized and specialized care settings. Significantly worse outcomes were observed among patients who shifted from generalized care settings to specialized settings (adjusted hazard ratio [aHR]=3.2, 95% confidence interval [CI]: 1.5, 6.8). In contrast, among the elderly, regular users of generalized care providers had a higher hazard of adverse outcomes than those of specialized care providers (aHR = 3.3, 95% CI: 1.0, 10.7). Conclusions: Diabetes patients who were persistently treated in generalized care settings had health outcomes comparable to those who were persistently treated in specialized care settings. However, for elderly and less stable patients, regular diabetes care in specialized care settings was preferable.