The purpose of this study was to identify factors that are associated with an inability to benefit from spinal manipulation as a treatment for low back pain. A total of 75 people with nonradicular low back pain participated. Subjects underwent a standardised examination that included history taking; self-reports of pain, disability and fear-avoidance beliefs; measurement of lumbar and hip range of motion; and use of various tests. All subjects received a spinal manipulation intervention for a maximum of two sessions. Subjects who did not show greater than five points of improvement on the modified Oswestry Low Back Pain Disability Questionnaire were considered to have shown no improvement with the manipulation. Baseline variables were tested for univariate relationship with the outcome of the manipulation. Variables showing a univariate relationship were entered into a logistic regression equation, and adjusted ORs were calculated. In all, 20 subjects (28%) did not improve after manipulation. Six variables were identified as being related to inability to improve with manipulation: longer symptom duration, having symptoms in the buttock or leg, absence of lumbar hypomobility, less hip rotation range of motion, less discrepancy in left-to-right hip medial rotation range of motion and a negative Gaenslen sign. The resulting logistic regression model explained 63% of the variance in manipulation outcome.