Introduction Seasonal influenza vaccine (SIV) is recommended in France for people aged 65 years or older and those with clinical risk factors, including diabetes. The French National Health Insurance Fund (NHIF) can identity these individuals and send them a voucher so they can obtain the vaccine free of charge. Despite this, SIV coverage remains below the target of 75% and has progressively decreased since the 2009 pandemic. We aimed to identify temporal trajectories (T) over a 10-year period of SIV uptake (SIVU) among patients with diabetes and to describe their correlates. Methods We identified patients with diabetes in 2006 (n = 17,259) among a representative sample of French NHIF beneficiaries. We followed them from 2006 through 2015, using SIV reimbursement claims and group-based trajectory modeling to identify SIVU-T and drug reimbursement claims to assess diabetes severity and comorbidity status. A multinomial logistic regression allowed us to study characteristics associated with the SIVU-T. Results We found 6 SIVU-T: (1) “never” T (prevalence: 32%): people with quasi null SIVU probabilities (≤ 5%); (2) “late increasing” T (4%): SIVU probability varying from ≤ 10% before 2011/12 to 85% in 2015/16; (3) “early increasing” T (8%): probability varying from ≤ 10% in 2006/07 to ≥ 80% starting in 2009/10; (4) “regular” T (33%): probability always ≥ 95%; (5) “progressively non-vaccinated” T (14%): probability decreasing from 90% in 2006/07 to 20% in 2015/16; (6) “post-pandemic decreasing” T (9%): probability decreasing right after the 2009/10 season. Overall, compared to the “never” T group, people in all other trajectories had poorer health at inclusion and/or over the study period (severe diabetes, high comorbidity score), people in the “late increasing” T group were more likely to have received newly free vaccination vouchers and/or changed general practitioners (GPs) during follow-up, and those in the “early increasing” T more likely to have received newly free vaccination vouchers and to have been hospitalized for an influenza-like illness during the follow-up. Those with “regular” T were slightly older than those with “never” T and less likely to have been hospitalized due to diabetes; those with a “progressively non-vaccinated” T were oldest (mean age at inclusion = 80 ± 7 years versus 65 ± 14 in the study population). Finally, people with a “post-pandemic decreasing” T were more frequently women, more likely to have been hospitalized due to an influenza-like illness, and to have changed GPs during the follow-up. Conclusions SIVU behavior was stable in most people with diabetes over the study period: one third were vaccinated regularly and another third never. The latter were globally healthier than the other groups and may feel less vulnerable to it, despite their clinical risk. About 25% became less inclined toward vaccination in one of two different patterns: the “progressively non-vaccinated” T may reflect the patient and/or healthcare professional's doubts about the benefits of SIV after a certain age (in part due to immunoscenescence); those in the “post-pandemic decreasing” T may have lost confidence in SIV after controversies in France during the 2009 mass vaccination campaign against the pandemic. About 10% became more likely to be vaccinated during the study period; our results suggest that receiving free vouchers for the first time might have triggered or fostered this behavioral change. Changing GPs and hospitalization for an influenza-like illness were associated with both increasing and decreasing trajectories. These events may represent key opportunities to foster or prevent behavioral changes toward SIV. Further research is needed to better understand the chronology of these events and potential causal pathways. These results should help stakeholders to adapt public health interventions to specific subgroups.