Early identification of patients at high risk of progression could help with a personalised treatment strategy. Magnetic resonance imaging (MRI) measures have been proposed to predict long-term disability in multiple sclerosis (MS), but a reliable predictor that can be easily implemented clinically is still needed. Assess MRI measures during the first 5years of the MS disease course for the ability to predict progression at 10+ years. Eighty-two MS patients (53 females), with ≥10years of clinical follow-up and having two MRI scans, were included. Clinical data were obtained at baseline, follow-up and at ≥10years. White matter lesion (WML) counts and volumes, and four linear brain sizes were measured on T2/FLAIR 'Fluid-Attenuated-Inversion-Recovery' and T1-weighted images. Baseline and follow-up inter-caudate diameter (ICD) and third ventricular width (TVW) measures correlated positively with Expanded Disability Status Scale, ≥10 or more of WMLs showed a high sensitivity in predicting progression, at ≥10years. A steeper rate of lesion volume increase was observed in subjects converting to secondary progressive MS. The sensitivity and specificity of both ICD and TVW, to predict disability at ≥10years were 60% and 64%, respectively. Despite advances in brain imaging and computerised volumetric analysis, ICD and TVW remain relevant as they are simple, fast and have the potential in predicting long-term disability. However, in this study, despite the statistical significance of these measures, the clinical utility is still not reliable.