BACKGROUND CONTEXT: Gadolinium-enhanced magnetic resonance imaging (Gd-MRI) is often performed in the evaluation of patients with persistent sciatica after lumbar disc surgery. However, correlation between enhancement and clinical findings is debated, and limited data are available regarding the reliability of enhancement findings. PURPOSE: To evaluate the reliability of Gd-MRI findings and their correlation with clinical findings in patients with sciatica. STUDY DESIGN: Prospective observational evaluation of patients who were enrolled in a randomized trial with 1-year follow-up. PATIENTS SAMPLE: Patients with 6- to 12-week sciatica, who participated in a multicentre randomized clinical trial comparing an early surgery strategy with prolonged conservative care with surgery if needed. In total 204 patients underwent Gd-MRI at baseline and after 1 year. OUTCOME MEASURES: Patients were assessed by means of the Roland Disability Questionnaire (RDQ) for sciatica, visual analog scale (VAS) for leg pain, and patient-reported perceived recovery at 1 year. Kappa coefficients were used to assess interobserver reliability. METHODS: In total, 204 patients underwent Gd-MRI at baseline and after 1 year. Magnetic resonance imaging findings were correlated to the outcome measures using the Mann-Whitney U test for continuous data and Fisher exact tests for categorical data. RESULTS: Poor-to-moderate agreement was observed regarding Gd enhancement of the herniated disc and compressed nerve root (kappa<0.41), which was in contrast with excellent interobserver agreement of the disc level of the herniated disc and compressed nerve root (kappa>0.95). Of the 59 patients with an enhancing herniated disc at 1 year, 86% reported recovery compared with 100% of the 12 patients with nonenhancing herniated discs (p=.34). Of the 12 patients with enhancement of the most affected nerve root at 1 year, 83% reported recovery compared with 85% of the 192 patients with no enhancement (p=.69). Patients with and without enhancing herniated discs or nerve roots at 1 year reported comparable outcomes on RDQ and VAS-leg pain. CONCLUSIONS: Reliability of Gd-MRI findings was poor-to-moderate and no correlation was observed between enhancement and clinical findings at 1-year follow-up. (C) 2014 Elsevier Inc. All rights reserved.