Objectives To investigate the association between insulin resistance (IR) and all-cause mortality among adults with rheumatoid arthritis (RA). Methods In this cohort study, we included 1,427 adults with self-reported RA from the National Health and Nutrition Examination Survey (NHANES) 1999–2010 and 2015–2018. IR was assessed using the homeostatic model assessment of insulin resistance (HOMA-IR). Survey-weighted multivariable Cox proportional hazards models were used to evaluate the association between HOMA-IR and all-cause mortality. Results During a mean follow-up of 9.7 years, 491 deaths occurred. Compared with participants in the lowest tertile of HOMA-IR, the multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause mortality were 0.633 (0.466–0.860) and 0.626 (0.463–0.847) for the second and third tertiles, respectively. Each 1-standard deviation increase in HOMA-IR was associated with a 14.5% lower risk of all-cause mortality (HR 0.855, 95% CI 0.734–0.995). The association varied by age and sex. Stratified analyses suggested that higher HOMA-IR was associated with increased mortality risk in men aged < 70 years and women aged < 50 years, whereas inverse associations were observed in men aged ≥ 70 years and women aged ≥ 50 years. Conclusions The association between HOMA-IR and all-cause mortality among adults with RA differed according to age and sex. Although higher HOMA-IR was inversely associated with mortality overall, subgroup-specific associations were heterogeneous and should not be interpreted as evidence of a protective effect of IR. Further studies are needed to clarify the mechanisms underlying these age- and sex-related differences. Keypoints • This study provides the first nationally representative evidence on the association between HOMA-IR and all-cause mortality among U.S. adults reporting rheumatoid arthritis. • HOMA-IR showed an approximately L-shaped association with all-cause mortality. • The association differed by age and sex, with higher mortality risk in younger subgroups and inverse associations in older subgroups. • The findings remained consistent across multiple sensitivity analyses addressing medication use, C-reactive protein (CRP) assessment, missing data, and potential reverse causation.