OBJECTIVES: To test the hypothesis that physical exercise induces an antiinflammatory response that is associated with reduced chronic activation of the tumor necrosis factor (TNF)-alpha system in frail elders and that the increase in muscle strength after resistance training is limited by systemic low-grade inflammation. DESIGN: A 12-week controlled resistance-training study. SETTING: Nursing homes in Copenhagen, Denmark. PARTICIPANTS: Twenty-one frail nursing home residents aged 86 to 95 completed the study. INTERVENTION: Ten participants were randomized to a program of resistance training of knee extensors and flexors three times a week for 12 weeks; the remaining 11 participants served as a control group who joined social activities supervised by an occupation therapist. MEASUREMENTS: Muscle strength, plasma levels of TNF-alpha, soluble TNF receptor (sTNFR)-1, and interleukin (IL)-6 were measured before and at the end of the intervention period. RESULTS: The training program improved muscle strength but did not affect plasma levels of TNF-alpha and sTNFR-I or IL-6. However, plasma levels of sTNFR-I at baseline were inversely correlated with the increase in muscle strength. CONCLUSION: Low-grade activation of the TNF system could limit the increase in muscle strength after resistance training in the oldest old. Furthermore, data suggest that the antiinflammatory response induced by 12 weeks of resistance training is not sufficient to reduce chronic activation of the TNF system, but the small sample size limited this interpretation.