Introduction: Recent evidence suggests that bone marrow lesions (BMLs) play a pivotal role in knee osteoarthritis (OA). The aims of this study were to determine: 1) whether baseline BML presence and/or severity predict sitespecific cartilage defect progression and cartilage volume loss; and 2) whether baseline cartilage defects predict site-specific BML progression. Methods: A total of 405 subjects (mean age 63 years, range 52 to 79) were measured at baseline and approximately 2.7 years later. Magnetic resonance imaging (MRI) of the right knee was performed to measure knee cartilage volume, cartilage defects (0 to 4), and BMLs (0 to 3) at the medial tibial (MT), medial femoral (MF), lateral tibial (LT), and lateral femoral (LF) sites. Logistic regression and generalized estimating equations were used to examine the relationship between BMLs and cartilage defects and cartilage volume loss. Results: At all four sites, baseline BML presence predicted defect progression (odds ratio (OR) 2.4 to 6.4, all P < 0.05), and cartilage volume loss (-0.9 to -2.9% difference per annum, all P < 0.05) at the same site. In multivariable analysis, there was a significant relationship between BML severity and defect progression at all four sites (OR 1.8 to 3.2, all P < 0.05) and BML severity and cartilage volume loss at the MF, LT, and LF sites (b -22.1 to -42.0, all P < 0.05). Additionally, baseline defect severity predicted BML progression at the MT and LF sites (OR 3.3 to 3.7, all P < 0.01). Lastly, there was a greater increase in cartilage volume loss at the MT and LT sites when both larger defects and BMLs were present at baseline (all P < 0.05). Conclusions: Baseline BMLs predicted site-specific defect progression and cartilage volume loss in a dose-response manner suggesting BMLs may have a local effect on cartilage homeostasis. Baseline defects predicted site-specific BML progression, which may represent increased bone loading adjacent to defects. These results suggest BMLs and defects are interconnected and play key roles in knee cartilage volume loss; thus, both should be considered targets for intervention.