This study investigated causes of attenuation of cerebrospinal fluid (CSF) signal on heavily T2-weighted (T2W) images in dogs with thoracolumbar disc extrusion. Medical records and magnetic resonance images were retrospectively reviewed. Dogs were classified into the following grades; grade 1, non-ambulatory paraparesis; grade 2, paraplegia with deep pain perception and grade 3, paraplegia without deep pain perception. The length of intramedullary T2W hyperintensity of the spinal cord, cranial/ caudal expansion of extradural compressive materials (ECM), and the CSF signal attenuation were measured. Ratios to the second lumbar vertebra (L2) were calculated for the length of intramedullary T2W hyperintensity (T2W:L2), cranial/caudal expansion of ECM (ECML:L2), and CSF signal attenuation (CSF:L2). The dogs were classified into focal or extended T2W hyperintensity groups according to the length [focal, shorter than length of L2; extended, longer than L2]. The area of EMC and the spinal canal were measured on transverse images at the lesion deriving occupancy ratio. The correlation between CSF:L2 and other data were analysed, and CSF:L2 was compared between the grades. In dogs with intramedullary T2W hyperintensity, the locations of CSF attenuation and the hyperintensity were compared if those locations were matched. Fifty-five dogs were included, 36 of which showed intramedullary T2W hyperintensity. Twenty-two of 36 dogs were considered as match of the location of the CSF attenuation and hyperintensity. CSF:L2 was significantly correlated with T2W:L2 in dogs with extended T2W hyperintensity (p = 0.0002), while CSF:L2 was significantly correlated with ECML:L2 in dogs with focal or no T2W hyperintensity (p = 0.0103 and p = 0.0364, respectively). CSF:L2 in grade 3 was significantly greater than those in patients who were grade 1 or 2 (both p < 0.001). In conclusion, higher CSF:L2, which was frequently seen in grade 3, would be most consistent with a higher T2W:L2 which might indicate spinal cord swelling.