Abstract The design of structures depends on the performance requirements that are usually defined through structural codes for single units. This approach may not be optimal because major earthquakes affect the built environment and can also trigger domino effects due to fires, explosions or toxic dispersion. To overcome this issue, we introduce a probabilistic framework for defining unit performance requirements by means of risk‐targeted fragility functions that account for potential domino effects and an overall system performance requirement. The system performance requirement is demonstrated by the potential loss of life, which represents the tolerated number of fatalities per year expected in the earthquake‐affected area. The proposed framework decomposes the problem into a damage‐based fatality analysis, population analysis and seismic hazard analysis, which can be performed and improved independently by engineers with the relevant expertise. The damage‐based fatality analysis includes a series of Monte Carlo simulations using logic trees to account for domino effects. These analyses are then coupled with an iterative risk estimation aimed at identifying the critical units in a system and estimating their risk‐targeted fragility functions, which can then be used for the design of a single unit. The methodology is demonstrated by estimating the potential loss of life due to a seismic event in an urban industrial area. It is shown that the risk‐targeted fragility functions of hazardous storage tanks can be estimated in a few iterations based on disaggregating the potential loss of life. The pilot research presented in the paper provides new possibilities for defining seismic design parameters of structures.