Recent reports postulate that the concomitant vertical deviation found in congenital superior oblique palsy is due to mechanical abnormalities rather than a congenitally paretic muscle, and is overcome in most patients by fusion. On the basis of the clinical characteristics alone a primary paresis is indeed unlikely. Although intraoperatively a different elasticity of the superior oblique tendon exists in congenital versus acquired cases of superior oblique palsy, preoperatively performed MR imaging shows that the clinical findings in congenital superior oblique muscle malfunction could nevertheless be of paretic origin. Seventeen consecutive patients (males: n = 13; females: n = 4) were examined. The vertical deviation in adduction was concomitant in vertical versions, the excyclotropia was small and concomitant in all directions of gaze and was less than 10 degrees even after diagnostic occlusion. All patients showed a positive Bielschowsky head tilt phenomenon and large fusional ability. We performed preoperative MR imaging of both orbits in high resolution 3 mm sections in coronal and axial orientations with and without contrast enhancement. In sixteen patients we found a significant reduction in muscle volume or even total aplasia of the superior oblique muscle of the affected side in comparison to the sound muscle on the other side. In contrast, two patients had a full blown clinical picture of a congenital superior oblique palsy but showed symmetrical muscle volumes on both sides in all coronal sections. Hypoplasia or aplasia of the superior oblique muscle on magnetic resonance imaging provides evidence for a primary paretic cause for the vertical squint found with congenital superior oblique dysfunction. It is not clear, however, whether this is caused by a primary hypoplasia or is of neurogenic origin. Our data together with the consistent difference in tendon morphology of the congenital and acquired forms of superior oblique palsy seem to exclude a purely neurogenic cause for the affection.