Duane retraction syndrome (DRS) is known to relate to the absence of the abducens nucleus, with abnormal innervation of the lateral rectus (LR) muscle by branchesof the oculomotor nerve (CN III). The purposes of this study were to investigate the morphological characteristics of the oculomotor nerve (CN III), the abducens nerve (CN VI), and the extraocular muscles in patients with clinically diagnosed Duane retraction syndrome (DRS) using MRI. In addition, we assessed the association between ocular motility, horizontal rectus muscle volumes, and CN III/VI in patients with Duane retraction syndrome (DRS). The study comprised 20 orthotropic control subjects (40 eyes) and 42 patients with Duane syndrome (48 eyes), including 20 patients with DRS Type I (24 eyes), 5 patients with DRS Type II (6 eyes), and 17 patients with DRS Type III (18 eyes). Three-dimensional (3D) T1/2 images of the brainstem and orbit were obtained to visualize the cranial nerves, especially the abducens (VI) and oculomotor (III) nerves, as well as extraocular muscles. Based on the clinical classification, among 42 patients, MRI showed that the abducens nerves (CN VI) on the affected side were absent in 24 of 24 eyes (100%; 20 patients) with Type I DRS and in 16 of 18 eyes (88%; 16 patients) with Type III DRS. However, CN VI was observed in 6 of 6 eyes (100%; 5 patients) with Type II DRS and in 2 of 18 eyes (11%) with Type III DRS. CN III was observed in all patients. The oculomotor nerves on the affected side were thicker than those on the nonaffected contralateral side in DRS Type I ( P < 0.05) and Type III ( P < 0.05), but not in DRS Type II. Smaller LR and larger MR volumes were shown in the affected eye than that in the nonaffected eye in DRS Types I and III. Based on the presence or absence of CN VI, there was a tendency for thicker oculomotor nerves in the affected eye than in the nonaffected eye in the absence groups ( P < 0.05). However, no significant difference was found in the present group. In the CN VI absence groups, similar results were found in the affected eyes than in the nonaffected eyes as in DRS Types I and III. In addition, the presence of CN VI was correlated with better abduction ( P = 0.008). The LR and MR volumes have positive correlations with the oculomotor nerve diameter in the affected eye. However, there was no correlation between the range of adduction/abduction and the LR/MR ratio in patients with or without an abducens nerve. Different types of DRS have different characteristic appearances of CN VI and CN III on MRI. Horizontal rectus muscles have morphological changes to adapt to dysinnervation of CN VI and aberrant innervation of CN III. Thus, these neuroimaging findings may provide a new diagnostic criterion for the classification of DRS, improving the comprehension of the physiopathogenics of this disease.