Background: The evidence for Internet-delivered pain management programs for chronic pain is growing, but there is little empirical understanding of how they effect change. Understanding mechanisms of clinical response to these programs could inform their effective development and delivery. Methods: A large sample (n = 396) from a previous randomized controlled trial of a validated internet-delivered psychological pain management program, the Pain Course, was used to examine the influence of three potential psychological mechanisms (pain acceptance, pain self-efficacy, fear of movement/re-injury) on treatment-related change in disability, depression, anxiety and average pain. Analyses involved generalized estimating equation models for clinical outcomes that adjusted for co-occurring change in psychological variables. This was paired with cross-lagged analysis to assess for evidence of causality. Analyses involved two time points, pre-treatment and post-treatment. Results: Changes in pain-acceptance were strongly associated with changes in three (depression, anxiety and average pain) of the four clinical outcomes. Changes in self-efficacy were also strongly associated with two (anxiety and average pain) clinical outcomes. These findings suggest that participants were unlikely to improve in these clinical outcomes without also experiencing increases in their pain self-efficacy and pain acceptance. However, there was no clear evidence from cross-lagged analyses to currently support these psychological variables as direct mechanisms of clinical improvements. There was only statistical evidence to suggest higher levels of self-efficacy moderated improvements in depression. Conclusions: The findings suggest that, while clinical improvements are closely associated with improvements in pain acceptance and self-efficacy, these psychological variables may not drive the treatment effects observed. Significance: This study employed robust statistical techniques to assess the psychological mechanisms of an established internet-delivered pain management program. While clinical improvements (e.g. depression, anxiety, pain) were closely associated with improvements in psychological variables (e.g. pain self-efficacy and pain acceptance), these variables do not appear to be treatment mechanisms.