Physiotherapists often use within-session changes to provide a guide for refining treatment application. This study tested the validity of within-session changes as predictors of between-session changes for patients with neck pain receiving manual therapy treatment. A total of 70 pairs of treatments from 29 patients with sub-acute non-specific neck pain receiving manual therapy were assessed to determine the relationship between within-session and between-session changes in range of motion (ROM), pain intensity, and centralisation. Measurements were taken of ROM of the more limited direction on each axis of flexion, extension, lateral-flexion and rotation, and pain (intensity and location) before and after treatment. The same measurements were repeated before the following treatment. Regression analysis demonstrated that within-session change accounted for 26% to 48% of the variability in between-session change for ROM and six per cent for pain intensity. The proportion of the within-session change for ROM maintained between sessions ranged from 42% to 63% (95% CI 25% to 88%). The odds ratios for within-session improved/not improved categorisation to predict between-session category for ROM ranged from 2.5 (95% CI 0.6 to 4.3) to 21.3 (95% CI 10.1 to 96.1), for pain intensity 4.5 (95% CI 1.2 to 14.4) and for pain centralisation 9.2 (95% CI 2.2 to 38.7) indicating greater likelihood of between-session improvement after within-session improvement. The between-session results for most patients (71% to 83%) could be classified correctly by their within-session category. The results support the use of within-session changes in ROM, centralisation, and possibly pain intensity as predictors of between-session changes for musculoskeletal disorders of the cervical spine.