Physiotherapists commonly use post-treatment changes in a patient's pain intensity and range of motion to guide treatment selection and predict possible longer-term outcomes. This study tested the validity of this practice by evaluating the predictive value of within-session changes in pain intensity and range of motion in 53 patients with low back pain. Pain intensity and range of motion measurements of spinal flexion, extension, lateral flexion, and straight-leg-raise were taken by the patient's therapist before and after one treatment session, and were repeated by a blinded therapist at the beginning of the patient's subsequent treatment session. Regression analysis revealed that the strength of association between within-session and between-session changes ranged from r = 0.35 to r = 0.80 for range of motion measurements, and from r = 0.24 to r = 0.47 for pain intensity. Odds ratios for pain and range of motion ranged from 3.5 (95% CI 0.9 to 14.6) to 37.0 (95% CI 4.1 to 330), indicating greater odds of improving between-session if improvement was obtained within-session. These results provide preliminary support for the practice of using within-session changes in pain intensity and range of motion to guide treatment selection when treating impairments in patients with low back pain.