Purpose: To assess the diagnostic performance of signal changes in Hoffa's fat pad (HFP) assessed on non-contrast-enhanced (CE) magnetic resonance imaging (MRI) in detecting synovitis, and the association of pain with signal changes in HFP on non-CE MRI and peripatellar synovial thickness on CE MRI. Methods: The Multicenter Osteoarthritis (MOST) Study is an observational study of individuals who have or are at high risk for knee OA. All subjects with available non-CE and CE MRIs were included. Signal changes in HFP were scored from 0 to 3 in two regions using non-CE MRI. Synovial thickness was scored from 0 to 2 on CE MRI in five peripatellar regions. Sensitivity, specificity and accuracy of HFP signal changes were calculated considering synovial thickness on CE MRI as the reference standard. We used logistic regression to assess the associations of HFP changes (non-CE MRI) and synovial thickness (CE MRI) with pain from walking up or down stairs, after adjusting for potential confounders. Results: A total of 393 subjects were included. Sensitivity of infrapatellar and intercondylar signal changes in HFP was high (71% and 88%), but specificity was low (55% and 30%). No significant associations were found between HFP changes on non-CE MRI and pain. Grade 2 synovial thickness assessed on CE MRI was significantly associated with pain after adjustments for potential confounders. Conclusion: Signal changes in HFP detected on non-CE MRI are a sensitive but non-specific surrogate for the assessment of synovitis. CE MRI identifies associations with pain better than non-CE MRI. © 2013 Osteoarthritis Research Society International.