Background context The clinical importance of lumbar pathology identified on magnetic resonance imaging (MRI) remains unclear. It is plausible that pathology seen on MRI is a risk factor for a recurrence of low back pain (LBP); however, to our knowledge, this has not been investigated by previous studies. Purpose The aim was to investigate whether lumbar pathology, identifiable on MRI, increases the risk of a recurrence of LBP. Study design This was a prospective inception cohort study with 1-year follow-up. Patient sample Seventy-six people who had recovered from an episode of LBP within the previous 3 months were included. Outcome measures The primary outcome was time to recurrence of LBP, which was determined by contacting participants at 2-month intervals for 12 months. Methods All participants underwent a baseline assessment including MRI scan and completion of a questionnaire, which assessed a range of potential risk factors for recurrence. Magnetic resonance imaging scans were reported for the presence of a range of MRI findings. The primary analysis investigated the predictive value of two clinical features (age and number of previous episodes) and six MRI findings (disc degeneration, high intensity zone, Modic changes, disc herniation, facet joint arthrosis, and spondylolisthesis) in a multivariate Cox regression model. We decided a priori that dichotomous predictors with hazard ratios (HRs) of greater than 1.5 or less than 0.67 would be considered potentially clinically important and justify further investigation. Results Of the eight predictors entered into the primary multivariate model, three (disc degeneration, high intensity zone, and number of previous episodes) met our a priori threshold for potential importance. Participants with disc degeneration score greater than or equal to 3 (Pfirrmann scale) had a HR of 1.89 (95% confidence interval [CI] 0.42-8.53) compared with those without. Patients with high intensity zone had an HR of 1.84 (95% CI 0.94-3.59) compared with those without. For every additional previous episode, participants had an HR of 1.04 (95% CI 1.02-1.07). Conclusions We identified promising risk factors for a recurrence of LBP, which should be further investigated in larger trials. The findings suggest that pathology seen on MRI plays a potentially important role in recurrence of LBP.