A recently developed near-infrared fluorescence-labeled folate probe (NIR2-folate) was tested for in vivo imaging of arthritis using a lipopolysaccharide intra-articular injection model and a KRN transgenic mice serum induction mouse model. In the lipopolysaccharide injection model, the fluorescence signal intensity of NIR2-folate (n = 12) and of free NIR2 (n = 5) was compared between lipopolysaccharide-treated and control joints. The fluorescence signal intensity of the NIR2-folate probe at the inflammatory joints was found to be significantly higher than the control normal joints (up to 2.3-fold, P < 0.001). The NIR2-free dye injection group showed a persistent lower enhancement ratio than the NIR2-folate probe injection group. Excessive folic acid was also given to demonstrate a competitive effect with the NIR2-folate. In the KRN serum transfer model (n = 4), NIR2-folate was applied at different time points after serum transfer, and the inflamed joints could be detected as early as 30 hours after arthritogenic antibody transfer (1.8-fold increase in signal intensity). Fluorescence microscopy, histology, and immunohistochemistry validated the optical imaging results. We conclude that in vivo arthritis detection was feasible using a folate-targeted near-infrared fluorescence probe. This receptor-targeted imaging method may facilitate improved arthritis diagnosis and early assessment of the disease progress by providing an in vivo characterization of active macrophage status in inflammatory joint diseases.