Introduction: Given the emerging evidence that osteoarthritis (OA) may have a vascular basis, the aim of this study was to determine whether serum lipids were associated with change in knee cartilage, presence of bone marrow lesions (BMLs) at baseline and the development of new BMLs over a 2-year period in a population of pain-free women in mid-life.Methods: One hundred forty-eight women 40 to 67 years old underwent magnetic resonance imaging (MRI) of their dominant knee at baseline and 2.2 (standard deviation 0.12) years later. Cartilage volume and BMLs were determined for both time points. Serum lipids were measured from a single-morning fasting blood test approximately 1.5 years prior to the MRI.Results: The incidence of BML at follow-up was associated with higher levels of total cholesterol (odds ratio [OR] 1.84, 95% confidence interval [CI] 1.01, 3.36; P = 0.048) and triglycerides (OR 8.4, 95% CI 1.63, 43.43; P = 0.01), but not high-density lipoprotein (HDL) (P = 0.93), low-density lipoprotein (LDL) (P = 0.20) or total cholesterol/HDL ratio (P = 0.17). No association between total cholesterol, triglycerides, HDL, LDL or total cholesterol/HDL ratio and presence of BMLs at baseline or annual change in total tibial cartilage volume was observed.Conclusions: In this study of asymptomatic middle-aged women with no clinical knee OA, serum cholesterol and triglyceride levels were associated with the incidence of BMLs over 2 years. This provides support for the hypothesis that vascular pathology may have a role in the pathogenesis of knee OA. Further work is warranted to clarify this and whether treatments aimed at reducing serum lipids may have a role in reducing the burden of knee OA.