Findings are presented from selective studies of injury, health care utilization, and health service response patterns in the Loma Prieta earthquake. Our overall purpose is to explore these areas as specific demand and supply factors useful in developing earthquake casualty estimation models. Specifically, the research reported here includes a broad analysis of work-related injuries and illnesses from the entire affected region, including where the injured sought care, and a case study of emergency medical response in heavily damaged Santa Cruz County. On the demand side, the injury pattern analysis found that: (1) almost all fatalities and most serious injuries were associated with structural failure; (2) the majority of fatalities and serious casualties came from a small number of damaged structures, most notably the Cypress Viaduct collapse; (3) falling sections of URM buildings, striking passersby and falling on top of other buildings, caused deaths and serious injuries; (4) while interior nonstructural elements and building contents posed a relatively low risk to building occupants, certain elements and settings were particularly hazardous; and (5) occupant actions contributed to injuries—with many victims injured while attempting to take protective actions. On the supply side, the health care utilization analysis found that 60% of the victims were treated in nonhospital settings. The emergency medical response case study found that: (1) the Santa Cruz 911 emergency response system was severely taxed, but remained functional and responsive, with the initial heavy demand reaching a normal level by day two; (2) advanced life support ambulance transports were fewer than normal, possibly due to a drop off in motor vehicle accidents; (3) the Santa Cruz County medical system was able to accommodate the injury load because the health care system was extensive; (4) although initially disrupted the three community hospitals did not suffer severe structural damage; and (5) the demand for major medical care was light. The previous findings have three major implications for earthquake casualty modeling. First, structural damage potential remains the major predictor of serious earthquake casualties, but, because nonstructural performance remains a key source of minor injuries and a potential risk for major ones, we should evaluate and prioritize nonstructural hazards in terms of their relative risk. Second, casualty estimates based exclusively on hospital utilization data may greatly underestimate the true medical demand. Third, future casualty models should estimate the resiliency of the entire local health care system.