To determine the accuracy of the surgeon's perception versus postoperative MRI findings in assessing the effectiveness of spinal cord decompression achieved in dogs surgically treated for intervertebral disk extrusion (IVDE) and whether postoperative MRI findings were more likely to be associated with various outcomes. 68 dogs surgically treated for cervical or thoracolumbar IVDE. Data on clinical, neurologic, pre- and postoperative MRI, and intraoperative findings as well as outcomes and recovery times (6-month follow-up period) were prospectively collected and compared between various groups. 54 (79%) dogs had thoracolumbar IVDE, and 14 (21%) had cervical IVDE. Median degree of spinal cord compression as assessed on transverse T2-weighted MRI images was 45.6% before surgery and 8.8% after surgery. The correlation between surgeons' perception (n = 3) and postoperative MRI findings for the degree of spinal cord decompression achieved was only fair (κ = 0.40). Unsatisfactory spinal cord decompression as assessed via postoperative MRI was associated with severity of preoperative neurologic grade and preoperative compression, thoracolumbar (vs cervical) IVDE, and ventral (vs ventrolateral or dorsolateral) circumferential distribution of extruded material. Satisfactory (vs unsatisfactory) decompression as assessed via MRI was associated with a lower postoperative neurologic grade, greater likelihood of a successful outcome, and lower mean recovery time. Results suggested that for dogs surgically treated for IVDE, the surgeon's perception of adequate spinal cord decompression may be less reliable than postoperative MRI findings. Postoperative MRI appeared particularly useful for dogs with a severe preoperative neurologic status, severe preoperative spinal cord compression, and thoracolumbar IVDE.