Background and purpose: The clinico-radiological paradox in multiple sclerosis (MS) is well recognized, relevant and yet poorly understood. The suitability of an in vivo model for the clinico-radiological paradox was tested, using internuclear ophthalmoplegia (INO) and the medial longitudinal fasciculus (MLF). Methods: In this cross-sectional study lesions of the MLF were rated by an experienced MS neuroradiologist blinded to all other information. The presence of an INO was objectively determined by a validated infrared oculography protocol (DEMoNS). Clinical information, including the National Eye Institute Visual Function Questionnaire, was obtained. Results: This study included 202 patients with MS. The clinico-radiological paradox occurred in 50 patients (25%). This consisted of 45 patients having an INO without an MLF lesion and five patients with an MLF lesion but without an INO. The visual function overall score was related to the presence of an INO (p = 0.016), but not to MLF lesions seen on magnetic resonance imaging (MRI) (p = 0.207). A consensus list of potential causes for the clinico-radiological paradox was compiled and the MRI images were deposited in a repository. Conclusion: This study provides an objective and quantitative model to investigate the clinico-radiological paradox. Our data suggest that pathology of the MLF is more frequently detected and more clinically relevant by infrared oculography than by MLF lesion rating on MRI.