This planned MiSpEx-Network reanalysis was designed to derive a dose-response relationship under consideration of further effect modifiers in exercises on low back pain. One thousand four hundred eighty three intervention participants with low back pain (mean age, 40.9 years [SD 14 years]) performed stabilisation exercises (3 weeks supervised, 9 weeks self-administered). Patients reported pain intensity, disability, and disability days at baseline, 3 weeks, 12 weeks, and 6 months post randomisation. Exercise characteristics and effect modifiers were prospectively monitored. Beyond the comparison to the results of the control group, linear mixed models were calculated to determine a dose-response relationship. The interventions led to small but significantly larger symptom reductions than in the control group at each measurement. Longer durations of the intervention led to larger symptom reductions. Higher exercise frequencies were associated with a decrease in pain intensity: to train once more per week led to a mean decrease of 0.93 points in pain intensity [95% CI = -1.54 to -0.32]. Disability days were also impacted (estimate = -0.07 [-0.14 to 0.00]), but disability was not (-0.09 [-0.67 to 0.48]). Adding perturbation was superior to adding stretching or a behavioural module. The relationships were robust when additional effect modifiers were considered and against the control group' effects. The odds ratio for a clinically important effect with higher exercise frequencies decreased at 3 weeks (OR = 0.71 [0.618-0.813] for >2.5*week -1 ) and increased at 12 weeks (1.13 [1.006-1.270], >1.5*week -1 ). Using longer intervention durations, adding a perturbation component to the stabilisation trainings and using higher frequencies (up to a certain point) may lead to an even more beneficial response on exercise in patients with low back pain. Developing strategies to maintain a training frequency of at least 2 times per week may be relevant in stabilisation exercises to treat low back pain.