The metabolic organization of plant cells poses a severe challenge for the development of the predictive models that are required for the rational manipulation of plant metabolism. While constraints-based network analysis, metabolic flux analysis, kinetic modeling, and metabolic control analysis provide a powerful complementary set of theoretical and empirical approaches for analyzing the structure and performance of plant metabolic networks, these tools have not yet led to easy solutions in the quest for useful targets for plant metabolic engineering. The task is particularly daunting in relation to the central pathways of carbon metabolism, where the metabolic characterization of transgenic plants reveals a remarkably robust metabolic network. These investigations indicate that the network can often compensate for alterations in the amounts of enzymes through changes in the steady-state levels of pathway intermediates and the activation state of the enzymes. Moreover, investigations of transgenic plants have revealed numerous instances of effects that arise as a secondary consequence of the original enzymic modification or that arise in pathways that seem at first sight to be quite separate from the pathway that is being manipulated. While it is clear that our qualitative understanding of primary plant metabolism is sufficient to rationalize the response of the metabolic network to changes in expression of a specific enzyme, it is difficult to believe that most of the responses that have been observed could be predicted with any degree of certainty with the currently available models. To do so would require a complete, quantitative understanding of all the relevant interactions between the components of the metabolic network and much further work will be required to achieve this goal.