Objective: A recent consensus statement suggested ≥3 corner inflammatory (CIL) or several corner fatty lesions (CFL) as candidate criteria for a positive spine MRI in axial spondyloarthritis. We aimed to evaluate the diagnostic utility of these cut-offs in non-radiographic axial spondyloarthritis (nr-axSpA) and ankylosing spondylitis (AS). Methods: 130 consecutive back pain patients ≤50 years newly referred to 2 university clinics (cohorts A/B) were classified according to rheumatologist expert opinion based on clinical examination and pelvic radiographs as having nr-axSpA (n=50), AS (n=33), or non-specific back pain (NSBP) (n=47). Cohort A also had 20 age-matched healthy controls. Four blinded readers assessed spine MRIs using the standardized Canada-Denmark module. Readers recorded CIL and CFL in 23 discovertebral units. We tested the diagnostic utility (mean sensitivity/specificity over 4 readers) of cut-off values for spinal MRI lesions as proposed in the literature (≥2/≥3 CIL; ≥6 CFL), and for possible thresholds from ≥1 to ≥10 CIL and CFL, for nr-axSpA and AS patients in both cohorts. Results: None of the spinal thresholds ≥2/≥3 CIL and ≥6 CFL showed clinically relevant diagnostic utility (range for positive likelihood ratios 1.38-2.36) when comparing nr-axSpA versus NSBP patients. A threshold of ≥6 CIL had moderate to substantial diagnostic utility (positive likelihood ratio 13.26/6.74) in nr-axSpA, while ≥4 CIL showed small diagnostic utility (3.83/2.72) but specificities >0.90. Conclusions: No previously proposed candidate criteria for a positive spinal MRI showed clinically relevant diagnostic utility in nr-axSpA. These findings question definitions of a positive MRI in spondyloarthritis based on spine MRI alone. This article is protected by copyright. All rights reserved.