Since real hydrocarbon fuels are commonly complex multi-component mixtures, their exact composition is generally neither known nor standardized. This makes the definition of surrogate mixtures to be used in simulations a crucial task to accurately model the evaporation and combustion behavior. Amongst others, this includes the specification of a certain numbers of surrogate components, the definition of optimization targets representing the real fuel properties of interest, the formulation of an appropriate numerical model for the evaluation of these quantities and the utilization of a suitable optimization algorithm to obtain the optimal surrogate composition. Concerning these points, several combinations can lead to a non-optimal representation of the real fuel target properties by the defined surrogate. A well-known example is the uncertainty in the numerical models often used for calculating the distillation curves measured by means of the ASTM D86 standard. Another example is the consideration of the chemical interactions between species which can influence the modeled fuel properties. Against this background, the sensitivity of the surrogate definition towards the used optimization targets and the numerical models, used to describe them, is analyzed in the present study. This investigation is based on representative real fuel datasets delivered by the FA Porsche AG including both, vapor pressure and distillation curve data.