Traditionally, the intracellular fluxes of complex metabolic networks were quantified by isotopic-tracer experiments, but, owing to practical limitations, ‘metabolic-flux balancing’ is emerging as an alternative. This has become an important tool for the quantitative analysis of the physiology of microorganisms and mammalian cells. It has been successfully applied to finding potential sites for metabolic engineering, determining metabolic capabilities and designing optimal feeding strategies. However, it has the fundamental problem that metabolic networks, and cyclic metabolic pathways in particular, are underdetermined. The search for constraints that can be used to determine fluxes correctly for a range of different conditions is an exciting challenge.