Objectives To investigate the association between magnetic resonance imaging (MRI), macroscopic and histological assessments of synovitis in end-stage knee osteoarthritis (KOA). Methods Synovitis of end-stage osteoarthritic knees was assessed using non-contrast-enhanced (CE), contrast-enhanced magnetic resonance imaging (CE-MRI) and dynamic contrast-enhanced (DCE)-MRI prior to (TKR) and correlated with microscopic and macroscopic assessments of synovitis obtained intraoperatively. Multiple bivariate correlations were used with a pre-specified threshold of 0.70 for significance. Also, multiple regression analyses with different subsets of MRI-variables as explanatory variables and the histology score as outcome variable were performed with the intention to find MRI-variables that best explain the variance in histological synovitis (i.e., highest R2). A stepped approach was taken starting with basic characteristics and non-CE MRI-variables (model 1), after which CE-MRI-variables were added (model 2) with the final model also including DCE-MRI-variables (model 3). Results 39 patients (56.4% women, mean age 68 years, Kellgren–Lawrence (KL) grade 4) had complete MRI and histological data. Only the DCE-MRI variable MExNvoxel (surrogate of the volume and degree of synovitis) and the macroscopic score showed correlations above the pre-specified threshold for acceptance with histological inflammation. The maximum R2-value obtained in Model 1 was R2 = 0.39. In Model 2, where the CE-MRI-variables were added, the highest R2 = 0.52. In Model 3, a four-variable model consisting of the gender, one CE-MRI and two DCE-MRI-variables yielded a R2 = 0.71. Conclusion DCE-MRI is correlated with histological synovitis in end-stage KOA and the combination of CE and DCE-MRI may be a useful, non-invasive tool in characterising synovitis in KOA.