We examined the reliability of radiological findings in predicting segmental instability in 112 patients (56 men, 56 women) with a mean age of 66.5 years (27 to 84) who had degenerative disease of the lumbar spine. They underwent intra-operative biomechanical evaluation using a new measurement system. Biomechanical instability was defined as a segment with a neutral zone > 2 mm/N. Risk factor analysis to predict instability was performed on radiographs (range of segmental movement, disc height), MRI (Thompson grade, Modic type), and on the axial CT appearance of the facet (type, opening, vacuum and the presence of osteophytes, subchondral erosion, cysts and sclerosis) using multivariate logistic regression analysis with a forward stepwise procedure. The facet type was classified as sagittally orientated, coronally orientated, anisotropic or wrapped. Stepwise multivariate regression analysis revealed that facet opening was the strongest predictor for instability (odds ratio 5.022, p = 0.009) followed by spondylolisthesis, MRI grade and subchondral sclerosis. Forward stepwise multivariate logistic regression indicated that spondylolisthesis, MRI grade, facet opening and subchondral sclerosis of the facet were risk factors. Symptoms evaluated by the Short-Form 36 and visual analogue scale showed that patients with an unstable segment were in significantly more pain than those without. Furthermore, the surgical procedures determined using the intra-operative measurement system were effective, suggesting that segmental instability influences the symptoms of lumbar degenerative disease