Abstract Objective To examine cognitive and brain resilience factors on older adult cognition following total knee arthroplasty (TKA). Methods Participants were recruited through orthopedic clinics and screened for eligibility (i.e., age > 60 years, English as first language, diagnosis of knee osteoarthritis, intact activities of daily living). Data were collected before surgery, three weeks, three months, and one-year post-TKA (N = 67; mean(SD) age = 69.49(7.09) years; mean(SD) education = 15.62(2.83) years; 51% female; 90% non-Hispanic White (NHW), 10% non-Hispanic Black. Participants received baseline magnetic resonance imaging (3 T Siemens Verio). Magnetization-prepared rapid gradient-echo images were preprocessed. Machine learning predicted brain age (DeepBrainNet) was calculated for each participant. Linear mixed effects models examined longitudinal effects of word reading (Wide Range Achievement Test-III (WRAT-III)), brain age gap (predicted brain age minus chronological age), sex, ethnicity/race, pain interference (Brief Pain Inventory), and depression (Beck Depression Inventory-II) on cognitive composites (i.e., memory, working memory/inhibition, reasoning, processing speed, language). Results Memory, working memory/inhibition, reasoning, and processing speed had curvilinear trajectories (p’s < 0.05), declining post-TKA and subsequently returning to baseline. Word reading positively associated with memory (β = 0.065, SE = 0.016, p < 0.001), working memory/inhibition (β = 0.069, SE = 0.016, p < 0.001), reasoning (β = 0.099, SE = 0.021, p < 0.001), and processing speed (β = 0.050, SE = 0.015, p < 0.01). Language negatively associated with BAG (β = −0.041, SE = 0.017, p < 0.05). Women had higher processing speed (β = −0.285, SE = 0.135, p < 0.05), and NHW participants had higher memory (β = 0.479, SE = 0.224, p < 0.05). When pain interference was higher, processing speed (β = −0.003, SE = 0.002, p < 0.05) and language (β = −0.007, SE = 0.002, p < 0.01) were lower. Conclusions Cognitive and brain resilience measures may identify individuals protected against cognitive dysfunction post-surgically. Future research increasing representation in underrepresented groups is warranted.