This study aims to elucidate the association between MRI measurements of spine alterations and self-reported outcomes of pain and disability in individuals with non-specific low back pain, using a comprehensive perspective based on machine learning algorithm. 246 consecutive subjects were assessed. Pain severity in cervical, lumbar, and leg regions was determined using visual analogue scale, and functional disability was acquired by Oswestry Disability Index. Sagittal and axial MRI scans of the thoracolumbar spine were evaluated. Severity of disc degeneration, spinal canal stenosis, and presence of vertebral endplate lesions based on two different classification schemes involving the extent and the shape of defects were quantified at the levels from T12L1 to L5S1. The following parameters describing the lumbar region as a whole were calculated: maximum value along spinal levels, sum of values along levels, number of levels characterized by severe condition. The association with pain and disability was assessed by generalized multiple linear regression modelling. Disc degeneration was identified as a predictor of disability and partially of pain, whereas canal stenosis was found associated with changes in pain in the leg region. Partial correlation values ranged from 0.11 to 0.32. Endplate lesions did not show significant associations. A partial association between MRI measurements and self-reported outcomes of pain and disability was confirmed. Disc degeneration was the most correlated with the reported indexes, while canal stenosis mainly affected the pain levels in the leg region. The presence of endplate lesions did not demonstrate any significant relationships.