This work aimed to assess inter-rater reliability and agreement of a magnetic resonance imaging (MRI)-based Kellgren and Lawrence (K&L) grading for patellofemoral joint osteoarthritis (OA) and to validate it against the MRI Osteoarthritis Knee Score (MOAKS). MRI scans from people aged 45 to 75��years with chronic knee pain participating in a randomised clinical trial evaluating dietary supplements were utilised. Fifty participants were randomly selected and scored using the MRI-based K&L grading using axial and sagittal MRI scans. Raters conducted inter-rater reliability, blinded to clinical information, radiology reports and other rater results. Intra- and inter-rater reliability and agreement were evaluated using the intra-class correlation coefficient (ICC) and Cohen���s weighted kappa. There was a 2-week interval between the first and second readings for intra-rater reliability. Validity was assessed using the MOAKS and evaluated using Spearman���s correlation coefficient. Intra-rater reliability of the K&L system was excellent: ICC 0.91 (95% CI 0.82���0.95); weighted kappa (����=��0.69). Inter-rater reliability was high (ICC 0.88; 95% CI 0.79���0.93), while agreement between raters was moderate (����=��0.49���0.57). Validity analysis demonstrated a strong correlation between the total MOAKS features score and the K&L grading system (����=��0.62���0.67) but weak correlations when compared with individual MOAKS features (����=��0.19���0.61). The high reliability and good agreement show consistency in grading the severity of patellofemoral OA with the MRI-based K&L score. Our validity results suggest that the scale may be useful, particularly in the clinical environment. Future research should validate this method against clinical findings.