This prospective cohort study investigates the prognosis of patients with neuropathic low back-related leg pain (LBLP) consulting in UK primary care. Data from 511 patients were collected using standardised baseline clinical examinations (including MRI scan findings), self-report questionnaires at baseline, 4-months, 12-months and 3-years. Cases of possible neuropathic pain (NP) and persistent-NP were identified using either of two definitions: i) clinical diagnosis of sciatica, ii) self-report version of Leeds Assessment for Neurological Symptoms and Signs (s-LANSS). Mixed-effects models compared pain intensity (highest of mean leg or mean back pain (0-10 NRS)) over 3-years between persistent-NP vs non-persistent NP based on i) clinical diagnosis, ii) s-LANSS. Logistic regression examined associations between potential prognostic factors and persistent-NP at 4-months based on the two NP definitions. At 4-months, using both definitions: i) approximately 4 out of 10 patients had persistent-NP, ii) mean pain intensity was higher for patients with persistent-NP at all follow-up points compared to those without, iii) only pain self-efficacy was significantly associated with persistent-NP (s-LANSS: OR 0.98, sciatica: 0.98), but it did not predict cases of persistent-NP in either multivariable model. Based on factors routinely collected from self-report and clinical examination, it was not possible to predict persistent-NP in this population. PERSPECTIVE: This study provides evidence that neuropathic back-related leg pain in patients consulting in primary care is not always persistent. Patients with persistent neuropathic pain had worse outcomes than those without. Neither leg pain intensity, pain self-efficacy nor MRI scan findings predicted cases of persistent neuropathic pain in this patient population.Copyright © 2023. Published by Elsevier Inc.