Current approaches to so-called Performance Based structural fire Design (PBD) lack a clear, objective safety format, and it is unclear whether the wide variety of specific approaches used result in an ‘adequate’ level of safety. However, also prescriptive design solutions given in guidance documents and standards have an unspecified target safety level. Thus, establishing a safety format for performance based design requires establishing target safety levels from first principles, rather than by comparison against prescriptive requirements. By applying Lifetime Cost Optimization (LCO), optimum design solutions and associated safety levels can be explicitly determined, and thus LCO can potentially form the basis for establishing a clear safety format for performance based structural fire design. As available calculation tools do not allow for a detailed modelling of all possible failure modes for structures exposed to fire, the possibility for rigorous application of LCO is currently limited. In this paper a hypothesis is formulated proposing that strength-based (‘collapse’) failure probabilities in the LCO formulation can be substituted by failure probabilities based on a simplified calculable limit state like deflection. As long as the simplified limit state is (1) unequivocally conservative, and (2) results in a close approximation of the incalculable limit state, the resultant optimum design obtained through LCO will be associated with only a negligible increase in investment in safety compared to the true (but unknown) optimum design solution. The validity of this hypothesis has been tentatively confirmed through an explicit evaluation for a simple example case.