Objectives: To determine the correlations among pain, disability, lumbar muscle endurance and fear-avoidance behaviour in patients with chronic low back pain and to investigate the predictive power of fear-avoidance behaviour on improvement in treatment outcomes. Design: Longitudinal retrospective correlation study. Participants: 178 patients with chronic low back pain. Intervention: Back reconditioning exercises using standardized protocol. Outcome measures: Pain, disability and lumbar muscle endurance were measured at week 0, 6 and 12. Fear-avoidance behaviour was measured at week 0. Changes in pain and disability were measured at week 6 and 12. Results: Fair to moderate correlation was found between pain and disability ( r = 0.35-0.62, p < 0.01). Fear Avoidance Behaviour Questionaire ( FABQ) correlated with pain at a weak degree ( r = 0.23, p < 0.01). FABQ correlated with disability at a fair degree ( r = 0.30-0.42, p < 0.05). FABQ correlated inversely with long-term improvement in disability at a fair degree ( r = -0.35, p < 0.05). FABQ significantly predicted long-term reduction in disability, accounting for 12% of the variance in outcome. Conclusions: No strong correlations were found among pain, disability, lumbar muscle endurance and fear-avoidance behaviour in patients with chronic low back pain. Treatment, therefore, should aim to address as many related aspects including physical, psychosocial and behavioural aspects, as possible, in order to enhance treatment success. Fear avoidance behaviour significantly predicted long-term reduction in disability, accounting for 12% of the variance in outcome.