Prospective cohort, test-retest design. Directional preference (DP) with centralization (CEN) and DP without CEN are common pain-pattern responses assessed by Mechanical Diagnosis and Therapy (MDT). Although there is evidence that MDT can reduce pain and disability in the short term by treating the patient with direction-specific exercises concordant with the patient's DP, the mechanism responsible for this is unclear. To determine whether clinical signs of impaired spinal control improve immediately after eliciting a DP-with-CEN response or a DP-without-CEN response in patients with nonspecific low back pain. Participants underwent a standardized MDT assessment and were classified into the following pain-pattern subgroups: DP with CEN, DP without CEN, or no DP. Clinical signs of impaired spinal control were assessed pre-MDT assessment and post-MDT assessment by an independent examiner. Four spinal control tests were conducted: aberrant lumbar movements while bending forward, the active straight leg raise (ASLR) test, the Trendelenburg test, and the prone instability test. Differences in spinal control pre-MDT assessment and post-MDT assessment were calculated for the 3 pain-pattern subgroups and compared with chi-square tests. We hypothesized that a larger proportion of patients in the DP-with-CEN subgroup would exhibit improved spinal control than patients categorized as DP without CEN or no DP. Of 114 patients recruited, 51 patients (44.7%) were categorized as DP with CEN, 23 (20.2%) as DP without CEN, and 40 (35.1%) as no DP. Before MDT assessment, between 28.9% (Trendelenburg test) and 63.7% (ASLR test) of patients showed impaired spinal control. After MDT assessment, a larger proportion of patients in the DP-with-CEN subgroup (43%) showed improvement than those in the no-DP subgroup (7%) on aberrant lumbar movements (P = .02). Likewise, more patients in the DP-with-CEN subgroup (50%) improved on the ASLR test than those in the no-DP subgroup (8%, P<.01) or the DP-without-CEN subgroup (7%, P = .01). Changes in Trendelenburg test and prone instability test outcomes did not reach statistical significance. Immediately following MDT assessment, a larger proportion of patients with a DP-with-CEN pain pattern showed improvement in clinical signs of spinal control compared to patients with a DP-without-CEN or no-DP pain pattern. The current study was registered in the Dutch trial registry at http://www.trialregister.nl/trialreg/index.asp (NTR4246). Therapy, level 2b.