Spinal canal stenosis is often measured on anatomical magnetic resonance imaging (MRI) to estimate the degree of spinal cord compression. This study examined whether two quantitative measures of spinal canal stenosis taken from anatomical MRI are related to spinal cord white-matter integrity in patients with cervical spondylosis measured by diffusion tensor imaging (DTI). DTI and T2-weighted MRI of the cervical spinal cord were performed in 15 patients with cervical spondylosis and ten healthy control subjects of similar age. Severity of stenosis was calculated using Pavlov's ratio and the space-available-for-cord (SAC) technique. Patients had significantly lower Pavlov's ratios and SAC (C2-C3, C4-C5 and C6-C7), lower fractional anisotropy (FA; C2-C3 and C4-C5) and higher radial diffusivity (C2-C3, C4-C5 and C6-C7) than the controls. In patients, only Pavlov's ratio correlated with mean FA (R=0.66, P=0.008). Variations in Pavlov's ratio and FA also showed a similar pattern across cervical levels. Pavlov's ratio is a better predictor of spinal cord integrity than the SAC and, therefore, may be more relevant clinically for the evaluation of stenosis in patients with cervical spondylosis.