OBJECTIVE:To determine whether loss in thigh muscle strength prior to knee replacement is caused by reductions of muscle strength in the anatomic cross-sectional area or by reductions of specific strength. METHODS:All 100 of the participants in the Osteoarthritis Initiative who underwent knee replacement and whose medical records included data on thigh isometric muscle strength and magnetic resonance imaging (MRI) (58 women, and 42 men, mean ± SD age 65 ± 8 years, mean ± SD body mass index [BMI] 29 ± 5 kg/m2 ) were matched with a control (no knee replacement) for age, sex, height, BMI, and radiographic severity. Thigh muscle anatomic cross-sectional area was determined by MRI at the research visit before knee replacement (time 0) and 2 years before time 0 (time -2). Specific strength (strength/anatomic cross-sectional area) was calculated, and the measures were compared by conditional logistic regression (i.e., odds ratio [OR] per standard deviation). ORs adjusted for pain (ORadj ) and 95% confidence intervals (95% CIs) were also calculated. RESULTS:Knee replacement cases had significantly smaller extensor (but not flexor) anatomic cross-sectional areas than controls at time 0 (women, ORadj 1.89 [95% CI 1.05-3.90]; men, ORadj 2.22 [95% CI 1.04-4.76]), whereas no significant differences were found at time -2. Women who had knee replacement showed lower levels of extensor specific strength than controls at time 0 (OR 1.59 [95% CI 1.02-2.50]), although this difference was not observed in men and did not maintain significance after adjustment for pain (ORadj 1.22 [95% CI 0.71-2.08]). Female cases lost significantly more extensor specific strength between time -2 and time 0 than controls (ORadj 3.76 [95% CI 1.04-13.60]), whereas no significant differences were noted at time -2, or in men. CONCLUSION:Prior to knee replacement, a significant reduction in knee extensor strength appears to occur in women through 2 mechanisms: one driven by pain (loss of specific strength) and one independent of pain (loss of muscle anatomic cross-sectional area). Men who underwent knee replacement showed significantly reduced levels of extensor anatomic cross-sectional area, but not significantly lower strength or specific strength.