Objective To investigate cross-sectional and longitudinal associations between serum leptin levels and knee cartilage thickness in older adults. Methods A prospective cohort of 163 randomly selected subjects (mean 63 years, range 52-78, 46% women) was studied. Knee cartilage thickness at medial tibial, lateral tibial, femoral and patellar sites was determined using T1-weighted fat-suppressed MRI. Serum leptin levels were measured by radioimmunoassay. Radiographic osteoarthritis, body fat (%), trunk fat (%), weight and height were measured, and body mass index (BMI) was calculated. Results Cross-sectionally, serum levels of leptin were negatively associated with femoral (β: −0.013, 95% CI −0.022 to −0.003), medial tibial (β: −0.009, 95% CI −0.018 to −0.001), lateral tibial (β: −0.012, 95% CI −0.021 to −0.003) and patellar (β: −0.014, 95% CI −0.026 to −0.002) cartilage thickness after adjustment for covariates. Moreover, BMI, trunk fat and total body fat were negatively associated with cartilage thickness, and the significant associations disappeared after further adjustment for leptin. Longitudinally, both baseline leptin and change in leptin were associated with greater changes in medial tibial cartilage thickness (β: −0.004, 95% CI −0.007 to −0.001 and β: −0.009, 95% CI −0.018 to −0.001, respectively) in multivariable analyses. Conclusions Serum levels of leptin are independently and consistently associated with reduced cartilage thickness cross-sectionally and longitudinally. In addition, the associations between adiposity measures and cartilage thickness are mediated by leptin, suggesting leptin may play a key role in cartilage thinning.