Objective: In order to increase sensitivity to detect longitudinal change, recording of within-grade changes was introduced for cartilage morphology and bone marrow lesion (BML) assessment in semiquantitative magnetic resonance imaging (MRI) scoring of knee osteoarthritis (OA). The aim of this study was to examine the validity provided by within-grade scoring. Design: The Multicenter Osteoarthritis (MOST) study is a longitudinal study of subjects with or at risk of knee OA. Baseline and 30 months MRIs were read according to the modified Whole-Organ Magnetic Resonance Imaging Score (WORMS) system including within-grade changes for cartilage and BMLs. We tested the validity of within-grade changes by whether the 30-month changes in cartilage and BML assessment were predicted by baseline ipsi-compartmental meniscal damage and malalignment, factors known to affect cartilage loss and BMLs, using ordinal logistic regression. Results: 1867 Knees (from 1411 participants) were included. Severe medial meniscal damage predicted partial grade (adjusted odds ratio (aOR) 4.4, 95% confidence interval (95% CI) 2.2, 8.7) but not ≥full grade (aOR 1.3, 95% CI 0.8, 2.2) worsening of cartilage loss and predicted both, partial grade (aOR 9.6, 95% CI 3.6, 25.1) and ≥full grade (aOR 5.1, 95% CI 3.2, 8.2) worsening of BMLs. Severe, but not moderate, malalignment predicted ipsi-compartmental within-grade (medial cartilage damage: aOR 5.5, 95% CI 2.6, 11.6; medial worsening of BMLs: aOR 4.9, 95% CI 2.0, 12.3) but not full grade worsening of BMLs and cartilage damage. Conclusions: Within-grade changes in semiquantitative MRI assessment of cartilage and BMLs are valid and their use may increase the sensitivity of semiquantitative readings in detecting longitudinal changes in these structures. © 2012 Osteoarthritis Research Society International.