Abstract Introduction Bone marrow lesion (BML) size may be an important imaging biomarker for osteoarthritis-related clinical trials and reducing BML size may be an important therapeutic goal. However, data on the interrelationships between BML size, pain, and structural progression are inconsistent and rarely examined in the same cohort. Therefore, we evaluated the cross-sectional and longitudinal associations of BML volume with knee pain and joint space narrowing (JSN). Methods A BML volume assessment was performed on magnetic resonance images of the knee collected at the 24- and 48-month Osteoarthritis Initiative visits from a convenience sample of 404 participants in the progression cohort. During the same visits, knee pain was assessed with WOMAC pain scores and knee radiographs were acquired and scored for JSN. BML volume was summed to generate a total knee volume and an index tibiofemoral compartment volume (compartment with greater baseline JSN). Primary analyses included multiple linear regressions (outcome = pain, predictor = total knee BML volume) and logistic regressions (outcome = JSN, predictor = index tibiofemoral compartment BML volume). Results This sample was 49% female with a mean age of 63 (9.2 standard deviation (SD)) years, and 71% had radiographic osteoarthritis in the study knee. Larger baseline BMLs were associated with greater baseline knee pain (P = 0.01), the presence of JSN at baseline (odds ratio (OR) = 1.50, 95% confidence interval (CI) = 1.23 to 1.83), and JSN progression (OR = 1.27, 95%CI = 1.11 to 1.46). Changes in total knee BML volume had a positive association with changes in knee pain severity (P = 0.004) and this association may be driven by knees that were progressing from no or small baseline BMLs to larger BMLs. In contrast, we found no linear positive relationship between BML volume change and JSN progression. Instead, regression of medial tibiofemoral BML volume was associated with JSN progression compared to knees with no or minimal changes in BML volume (OR = 3.36, 95%CI = 1.55 to 7.28). However, follow-up analyses indicated that the association between JSN progression and BML volume change may primarily be influenced by baseline BML volume. Conclusion Large baseline BMLs are associated with greater baseline knee pain, the presence of JSN at baseline, and disease progression. Additionally, BML regression is associated with decreased knee pain but not a reduced risk of concurrent JSN progression.