Abstract: U.S. Service members and Veterans (SM/V) experience elevated rates of traumatic brain injury (TBI), chronic pain, and other non-pain symptoms. However, the role of non-pain factors on pain interference levels remains unclear among SM/Vs, particularly those with a history of TBI. The primary objective of this study was to identify factors that differentiate high/low pain interference given equivalent pain intensity among U.S. SM/V participating in the ongoing LIMBIC-CENC national multicenter prospective longitudinal observational study. An explainable machine learning was used to identify key predictors of pain interference conditioned on equivalent pain intensity. The final sample consisted of N=1,577 SM/V who were predominantly male (87%), and 83.6% had a history of mild TBI(s), while 16.4% were TBI negative controls. The sample was categorized according to pain interference level (Low: 19.9%, Moderate: 52.5%, and High: 27.6%). Both pain intensity scores and pain interference scores increased with number of mild TBIs (p<0.001), and there was evidence of a dose response between number of injuries and pain scores. Machine learning models identified fatigue and anxiety as the most important predictors of pain interference, while emotional control was protective. Partial dependence plots identified marginal effects of fatigue and anxiety were associated with pain interference (p<0.001), but the marginal effect of mild TBI was not significant in models considering all variables (p>0.05). Non-pain factors are associated with functional limitations and disability experience among SM/V with mild TBI history. The functional effects of pain may be mediated through multiple other factors. Pain is a multidimensional experience that may benefit most from holistic treatment approaches that target comorbidities and build supports that promote recovery.