Background. Efforts to identify specific variables that impact most on outcomes from interdisciplinary pain rehabilitation are challenged by the complexity of chronic pain. Methods to manage this complexity are needed. The purpose of the study was to determine the network structure entailed in a set of self-reported variables, examine change, and look at potential predictors of outcome, from a network perspective. Methods. In this study we apply network analysis to a large sample of people seeking interdisciplinary pain treatment (N = 2,241). Variables analyzed include pain intensity, pain interference, extent of pain, depression, anxiety, insomnia, and psychological variables from cognitive behavioral models of chronic pain. Results. We found that Acceptance, Pain Interference, and Depression were key, "central," variables in the pretreatment network. Interestingly, there were few changes in the overall network configuration following treatment, specifically with respect to which variables appear most central relative to each other. On the other hand, Catastrophizing, Depression, Anxiety, and Pain Interference each became less central over time. Changes in Life Control, Acceptance, and Anxiety were most strongly related to changes in the remainder of the network as a whole. Finally, no network differences were found between treatment responders and non-responders. Conclusions. This study highlights potential future targets for pain treatment. Further application of a network approach to interdisciplinary pain rehabilitation data is recommended. Going forward, it may be better to next do this in a more comprehensive theoretically guided fashion, and ideographically, to detect unique individual differences in potential treatment processes.