Objective Little is known about early knee osteoarthritis (OA). The significance of lesions on magnetic resonance imaging (MRI) in older persons without radiographic OA is unclear. Our objectives were to determine the extent of tissue pathology by MRI and evaluate its significance by testing the following hypotheses: cartilage damage, bone marrow lesions, and meniscal damage are associated with prevalent frequent knee symptoms and incident persistent symptoms; bone marrow lesions and meniscal damage are associated with incident tibiofemoral (TF) cartilage damage; and bone marrow lesions are associated with incident patellofemoral (PF) cartilage damage. Methods In a cohort study of 849 Osteoarthritis Initiative (OAI) participants who had a bilateral Kellgren/Lawrence (K/L) score of 0, we assessed cartilage damage, bone marrow lesions, and meniscal damage using the MRI OA Knee Score, as well as prevalent frequent knee symptoms, incident persistent symptoms, and incident cartilage damage. Multiple logistic regression (one knee per person) was used to evaluate associations between MRI lesions and each of these outcomes. Results Of the participants evaluated, 76% had cartilage damage, 61% had bone marrow lesions, 21% had meniscal tears, and 14% had meniscal extrusion. Cartilage damage (any; TF and PF), bone marrow lesions (any; TF and PF), meniscal extrusion, and body mass index (BMI) were associated with prevalent frequent symptoms. Cartilage damage (isolated PF; TF and PF), bone marrow lesions (any; isolated PF; TF and PF), meniscal tears, and BMI were associated with incident persistent symptoms. Hand OA, but no individual lesion type, was associated with incident TF cartilage damage, and bone marrow lesions (any; any PF) with incident PF damage. Having more lesion types was associated with a greater risk of outcomes. Conclusion MRI‐detected lesions are not incidental and may represent early disease in persons at increased risk of knee OA.