The opioid crisis has remained a major public health challenge in the United States for many years. This study develops a data-driven decision support framework to guide policymakers in allocating county-level budgets across multiple expenditure categories in order to address the opioid crisis. We compile and curate a detailed dataset on fiscal policy and opioid-related outcomes in West Virginia (WV), a state severely affected by the epidemic. Drawing on this data set, we employ a Fixed Effects (FE) model to estimate the relationships between county-level budget allocations and two outcomes associated with the opioid crisis: Opioid-Related Deaths (ORDs) and Crime Incidents (CIs). We then develop combined FE and Tree Ensemble (TEs) models, in which TEs capture remaining input-output relationships not explicitly represented in the linear FE model. We embed the resulting FE+TEs models within a Mixed Integer Programming (MIP) formulation that identifies budget allocations across expenditure categories to maximize policymakers’ worst-case, risk-averse utility over ORDs and CIs. We apply the framework to 12 counties in southern WV that have been most impacted by the opioid crisis. When the two outcomes are given equal emphasis, the recommended allocations reduce the predicted levels of both ORDs and CIs; however, the results also reveal trade-offs between the two outcomes when policymakers place greater emphasis on one relative to the other. Furthermore, sensitivity analysis with respect to the total available budget shows that additional resources do not necessarily improve predicted outcomes when the flexibility to reallocate those resources across expenditure categories is limited. More broadly, the results demonstrate how the proposed framework can generate tailored, county-level budget recommendations to assist policymakers in reprioritizing expenditures to address the opioid crisis.