Genome-scale metabolic networks can be reconstructed using a constraint-based modeling approach. The stoichiometry of the network and the physiochemical laws still enable organisms to achieve certain objectives -such as biomass composition- through many various pathways. This means that the system is under-determined and many alternative solutions exist. A recent popular method used to reduce the number of alternative pathways is Flux Balance Analysis (FBA), which tries to optimize a given biological objective function. FBA does not always find a correct solution and for many networks the biological objective function is simply unknown. As a consequence researchers may have to measure certain fluxes. In this talk we discuss a method that combines a sampling approach with a greedy algorithm for finding a small given number of fluxes that, if measured, are expected to reduce as much as possible the solution space towards the true flux distribution. The method can be used for guiding the biologists to perform experimental analysis of metabolic networks. We will also shortly review studies that focus on constraint-based models of normal and malfunctioning mitochondria.