A method for the prediction of the interactions within complex reaction networks from experimentally measured time series of the concentration of the species composing the system has been tested experimentally on the first few steps of the glycolytic pathway. The reconstituted reaction system, containing eight enzymes and 14 metabolic intermediates, was kept away from equilibrium in a continuous-flow, stirred-tank reactor. Input concentrations of adenosine monophosphate and citrate were externally varied over time, and their concentrations in the reactor and the response of eight other species were measured. Multidimensional scaling analysis and heuristic algorithms applied to two-species time-lagged correlation functions derived from the time series yielded a diagram from which the interactions among all of the species could be deduced. The diagram predicts essential features of the known reaction network in regard to chemical reactions and interactions among the measured species. The approach is applicable to many complex reaction systems.