The objective of this study is to empirically and systematically assess the combination of inherent susceptibility factors, protective actions, and factors of hazard exposure that influence a house-hold's level of hardship experienced due to disruptions in critical infrastructure services during disasters. Classification and regression tree (CART) decision tree models and survey data from three major hurricane events were used to: (1) identify the pathways leading to impact(s) due to service disruptions and explore the differences in pathways across vulnerable population groups; and (2) identify the points of intervention to mitigate well-being impacts in households due to disruptions in water, energy, food, and road transportation services. The results reveal how the associative pathways between these factors change between socioeconomic and demographic groups in the impacted community and for different infrastructure service system types. The findings suggest that not all vulnerable households experienced high hardship outcomes despite prolonged outages. Finally, the hardship pathways suggest recommendations for improving resilience in infrastructure systems in a more equitable manner. The findings can be used by emergency and infrastructure managers and operators to better prioritize resource allocation for hazard mitigation investments and restorations. Accordingly, this study contributes to the theory of human-centric infrastructure resilience.