Research in network optimizationhas reached the stage where large-scale problems-linear or nonlinear, pure or generalized-are solved very efficiently with minimal computing resources. Representing such problems for solution on the computer, however, remains a rather cumbersome task. Taking advantage of developments in high-level modeling languages, we design and implement integrated systems to facilitate the representation and solution of network problems. Such systems integrate the flexibility and robustness of modeling languages with the efficiency of network optimizers.We describe two alternative modes for this integration, which can be achieved for linear and nonlinear problems alike. The use of the resulting systems is demonstrated with the solution of largescale problems from diverse applications and with the implementation of network decomposition algorithms.