The choice of location is a key facet of decision-making in operations. One such choice is whether to colocate activities, services, or personnel. Prior research, including in healthcare, has reported that colocation yields benefits. However, these benefits may need to be balanced with higher costs of colocation (e.g., due to operational constraints). Thus, it is critical to understand not only whether colocation makes a difference but also under what circumstances it is most beneficial, and the mechanisms through which those benefits are realized. We consider colocation in the context of healthcare services, and ask: Does colocation of mental and physical health resources improve patient outcomes? This colocation is important, as primary care serves as a gateway to address mental health concerns and referrals to specialists. We next study colocation’s relationship with two important operational variables: continuity of care (CoC) with a provider, and patient severity. Finally, we examine the mediating role of patients’ no-shows and medication adherence in the colocation-outcomes relationship. As America’s largest integrated healthcare system, the Veterans Health Administration offers us an excellent setting to study these questions. We analyze over 300,000 patients – over an eleven-year period – who suffer from diabetes, a chronic condition, and show evidence of mental illness. We use an empirical approach to quantify the relationship between colocation and four key outcomes attributable to mental illness: hospitalizations, length of stay (LOS), 30-day readmissions, and suicidal behavior. We find that colocation is associated with improvement in outcomes. For example, a one standard deviation increase in the mean colocation measure is related to a 2.4% decrease in LOS – equivalent to an annual savings of approximately $1.5 million, on average, just for our cohort. In addition, we find that colocation and CoC are substitutes, in that colocation benefits patients whose care is fragmented. Further, we find that colocation offers greater benefits to patients whose mental health conditions are more severe. Finally, our analysis reveals that colocation improves outcomes (partially) through a reduction in the no-show rate and an increase in medication adherence. Our findings are validated by extensive robustness checks and sensitivity analyses. Our study has implications for both the theory and practice of healthcare operations. Theoretically, we advance the location literature, establish its connection with the continuity literature, and highlight key moderators and mediators in the colocation-outcomes relationship. Practically, our work offers insights into how to design an operationally efficient system and, in settings with limited resources, where to target colocation. Our study is particularly timely as it may help address the growing mental health crisis in America and around the world, further exacerbated during the COVID-19 pandemic.