The aim was to evaluate prevalence and factors associated with anti-tumor necrosis factor (anti-TNF) de-escalation in older adults with rheumatoid arthritis (RA). We identified adults ≥66 years of age with RA on anti-TNF therapy within 6 months after RA diagnosis with at least 6-7 months duration of use (proxy for stable use), using 20% Medicare data from 2008-2017. Patient demographic and clinical characteristics, including concomitant use of glucocorticoid (GC), were collected. Anti-TNF use was categorized as either de-escalation (identified by dosing interval increase, dose reduction, or cessation of use) or continuation. We used (1) an observational cohort design with Cox regression to assess patient characteristics associated with de-escalation and (2) a case-control design with propensity score-adjusted logistic regression to assess the association of de-escalation with different clinical conditions and concomitant medication use. We identified 5106 Medicare beneficiaries with RA on anti-TNF, 65.5% of whom had de-escalation. De-escalation was more likely with older age (hazard ratio [HR] 1.01, 95% confidence interval [CI] 1.01-1.02) or greater comorbidity (HR 1.07, 95% CI 1.05-1.09), but was less likely with low-income subsidy status (HR 0.85, 95% CI 0.78-0.92), adjusting for patient sex and race/ethnicity. Lower odds of de-escalation were associated with serious infection (odds ratio [OR] 0.79, 95% CI 0.66-0.94), new heart failure diagnosis (OR 0.70, 95% CI 0.52-0.95), and long-term GC use (OR 0.84, 95% CI 0.74-0.95), whereas higher odds were associated with concomitant methotrexate use (OR 1.16, 95% CI 1.03-1.31). Anti-TNFs are de-escalated in two-thirds of older adults with RA in usual care. Further study is needed on RA outcomes after anti-TNF de-escalation.