OBJECTIVE: To investigate the changes in 2 electromyographic measures, flexion relaxation (FR) response and feed-forward activation of the deep abdominals, associated with low back pain (LBP) after different rehabilitation interventions. DESIGN: A 2x2 factorial design with subjects' self-selecting treatment with randomization after 4 weeks to either the specific exercise group or exercise advice group for a further 12-week period. SETTING: General community practitioners and university training center. PARTICIPANTS: Subjects with chronic nonspecific LBP were recruited for this study. A total of 112 people were initially screened, and 60 were recruited for the study, with 50 being available for long-term follow-up. INTERVENTION: Four weeks of treatment (manipulative or nonmanipulation) and 12 weeks of subsequent exercise (supervised Swiss ball training or exercise advice). MAIN OUTCOME MEASURES: The Oswestry Disability Index, FR response measured at T12-L1 and L4-5, and feed-forward activation of the deep abdominal muscles. RESULTS: More rapid improvements in disability were identified for subjects who received the supervised exercise program. The FR response at L4-5 also increased more for those who received directly supervised exercise. Long-term follow-up showed that there was still a between-group difference in the FR response, despite no difference in self-rated disability. Long-term changes were observed for the feed-forward activation of the deep abdominals; however, no exercise or treatment effects were identified. CONCLUSIONS: Supervised exercise rehabilitation leads to more rapid improvements in self-rated disability, which were associated with greater improvement in the low back FR response.