Objectives: The aim of this study was to determine whether certain patient characteristics could identify people with acute low back pain who were more likely to respond to nonsteroidal anti-inflammatory drugs (NSAIDs), when administered in combination with paracetamol. Methods: This study involved a secondary analysis of a randomized controlled trial investigating the efficacy of diclofenac in 239 patients presenting to general practitioners for acute low back pain. All patients received advice to "stay active" and take regular paracetamol and then were randomized to receive either diclofenac (50mg twice daily) or placebo. The primary outcome was days to recovery from pain. The ability of 14 patient characteristics to identify those who respond best to diclofenac was assessed using interaction terms in Cox regression models. Results: Most of the 14 baseline characteristics investigated were uninformative in identifying those who respond best to diclofenac when added to paracetamol. Patients' sex and levels of fear regarding movement and pain were both independent statistically significant predictors of response to NSAIDs for one definition of recovery but not for the other. The interaction between fear avoidance (physical activity) and NSAIDs treatment was statistically significant (P=0.042, hazard ratio=1.059, 95% confidence interval 1.002 to 1.118) for recovery defined as a pain score of 0 or 1. The interaction between sex and NSAID treatment was statistically significant (P=0.044, hazard ratio=1.755, 95% confidence interval 1.014 to 3.038) for recovery defined as a pain score of 0 or 1 maintained for 7 consecutive days. Discussion: This study did not find any baseline patient characteristics, which consistently identified patients with acute low back pain who respond best to NSAIDs when provided in addition to paracetamol and advice. A patient's sex and level of fear avoidance may be weak predictors and warrant further prospective investigation.