Background context: Intramedullary signal intensity changes on magnetic resonance imaging (MRI) in cervical spondylotic myelopathy are thought to be indicative of the prognosis. However, the prognostic significance of signal intensity changes remains controversial. Purpose: To determine the radiographic and clinical factors that correlate with the prognosis after surgery in patients with cervical spondylotic myelopathy and to investigate the factors affecting the outcome of intramedullary signal changes on MRI. Study design: A prospective study evaluating clinical parameters and MRI in consecutive patients operated on for cervical spondylotic myelopathy. Patient sample: A total of 146 consecutive patients with cervical spondylotic myelopathy operated on during a 2-year period (September 1999 to September 2001) formed the study group. Outcome measures: Age, duration of symptoms, number of cervical prolapsed intervertebral discs, surgical approach, preoperative signal changes, residual compression and postoperative outcome of signal changes; clinical outcome (motor, sensory, autonomic and disability improvement). Methods: The participants in this study underwent anterior cervical discectomy/corpectomy or laminectomy/laminoplasty for cervical spondylotic myelopathy. Clinical features and MRI findings were studied in detail and compared with postoperative clinical and radiological status. The spinal cord signal intensity changes were evaluated before and after surgery. The multifactorial effect of such variables as age, duration of symptoms, number of prolapsed intervertebral discs, surgical approach (anterior/posterior), preoperative cord changes on T1- and T2-weighted sequences and persistence/regression of cord changes on clinical outcome (motor/sensory/autonomic/disability improvement) was studied using stepwise logistic regression. The highlight of the study is the analysis of the factors affecting regression of cord changes and their effect on postoperative outcome. Results: Preoperative intramedullary signal changes were present in 121 of 146 patients (82.9%); of these 121 patients, T1- and T2-weighted images were present in 81, and T2-weighted images were present in 40 (no patient had isolated T1 change). Postoperative MRI could be obtained in 44 of 121 patients (36.4%) with preoperative intramedullary signal changes; 14 had regression of cord changes. There was no significant difference in the clinical presentation of patients with and without cord changes. There was a significant correlation between the surgical outcome of patients and their age, duration of symptoms, number of cervical prolapsed intervertebral discs, surgical approach, preoperative signal changes, residual compression and postoperative outcome of signal changes. The patients with no intramedullary signal changes and signal changes only on T2-weighted images had a better outcome than patients with signal changes on both T1- and T2-weighted images. The patients with regression of intramedullary signal changes had significantly better outcome. There was no significant correlation between regression of signal changes and other factors. However, chronicity of disease, multiplicity of discs and postoperative residual compression relatively affect persistence of intramedullary signal changes. Conclusions: The presence of intramedullary signal changes on T1- as well as T2-weighted sequences on MRI in patients with cervical spondylotic myelopathy indicates a poor prognosis. However, the T2 signal intensity changes reflect a broad spectrum of spinal cord reparative potentials. Predictors of surgical outcomes are preoperative signal intensity change patterns of the spinal cord and their postoperative persistence/regression on radiological evaluations, age at the time of surgery, multiplicity of involvement and chronicity of the disease and surgical approach (anterior/posterior).