To describe prevalence of osteophytes (OPs) detected only by magnetic resonance imaging (MRI) but not by standard X-ray in older adults and to evaluate longitudinal associations with knee structural changes.837 participants were randomly selected from the local community and had MRI scans to assess knee OPs and other structures. OPs detected only by MRI but not by standard X-ray were defined as MRI-detected early OPs (MRI-OPs for short). OPs detected by both MRI and X-ray were defined as established-OPs.The prevalence of MRI-OPs was 50% while the prevalence of established-OPs was 10% and no-OPs was 40% at total tibiofemoral (TF) compartment at baseline. Compared with no-OPs, participants with MRI-OPs had greater risks of increased cartilage defects in all TF compartments (RR 1.37, 95%CI 1.07–1.74) and bone marrow lesions (BMLs) only in medial TF compartment (RR 1.49, 95%CI 1.06–2.11), after adjustment for age, sex, BMI, cartilage defects, BMLs and/or joint space narrowing; participants with established-OPs had greater cartilage volume loss at total (β −2.02, 95%CI −3.86, −0.17) and lateral tibial sites (β −5.63, 95%CI −9.93, −1.32), greater risks of increased cartilage defects in total (RR 1.66, 95%CI 1.15–2.40) and medial TF compartments (RR 1.49, 95%CI 1.20–1.69) and BMLs in all TF compartments (RR 1.88, 95%CI 1.22–2.89), after adjustment for covariates.MRI-OPs were associated with changes in knee structures, and the associations were similar but not as prominent as those for established-OPs. These suggest MRI-OPs may have a role to play in knee early-stage osteoarthritic progression.