Background: Understanding successful and unsuccessful behavioural treatment for pain is essential. Aims: We carried out a retrospective survey of 130 people who had undergone pain rehabilitation based on acceptance and commitment therapy, aiming to identify factors associated with non-response. Method: The sample was selected using the reliable change index to define ‘responders’ and ‘non-responders’ to key outcome measures. We surveyed a range of treatment-related, systemic, practical and personal factors that may have affected their treatment, and then compared ‘non-responders’ with ‘responders’, controlling for factors that might not be causal or specific to non-response. Results: Logistic regression analysis showed two themes that distinguished the groups, ‘people outside programme’ and ‘emotional state’. Conclusions: These data have clinical implications, as such factors can be addressed directly or incorporated into an assessment of treatment ‘readiness’. This study introduced a novel methodology for the investigation of pain treatment response, which allowed a broad study of clinically relevant variables, but with greater rigour than conventional self-reports of ‘helpful factors’ in treatment.