Metabolic networks of 43 organisms are organized into small, highly connected modules that combine hierarchically into larger units, with their number and clustering following a power law. In Escherichia coli, this hierarchical modularity aligns with known metabolic functions. The study shows that metabolic networks have a scale-free topology, characterized by a few highly connected nodes, but also exhibit a high clustering coefficient, suggesting modularity. This creates a conflict between scale-free and modular models. A hierarchical network model is proposed, which combines scale-free topology with inherent modularity. This model has a power law degree distribution and a high clustering coefficient, matching observations in metabolic networks. The hierarchical organization is supported by the scaling law C(k) ~ k^{-1}, indicating a hierarchy of nodes with different degrees of modularity. Analysis of E. coli's metabolic network shows that topological overlap between substrates correlates with their functional classification. Hierarchical clustering of the topological overlap matrix reveals distinct modules, which align with known functional classes. The study suggests that metabolic networks have a hierarchical modularity that reconciles their scale-free topology, high clustering coefficient, and power-law scaling of C(k). This hierarchical organization is consistent with the idea that evolution acts at multiple levels, affecting both small and large modules. The findings suggest that hierarchical modularity is a general feature of biological networks, applicable to other cellular and complex systems.Metabolic networks of 43 organisms are organized into small, highly connected modules that combine hierarchically into larger units, with their number and clustering following a power law. In Escherichia coli, this hierarchical modularity aligns with known metabolic functions. The study shows that metabolic networks have a scale-free topology, characterized by a few highly connected nodes, but also exhibit a high clustering coefficient, suggesting modularity. This creates a conflict between scale-free and modular models. A hierarchical network model is proposed, which combines scale-free topology with inherent modularity. This model has a power law degree distribution and a high clustering coefficient, matching observations in metabolic networks. The hierarchical organization is supported by the scaling law C(k) ~ k^{-1}, indicating a hierarchy of nodes with different degrees of modularity. Analysis of E. coli's metabolic network shows that topological overlap between substrates correlates with their functional classification. Hierarchical clustering of the topological overlap matrix reveals distinct modules, which align with known functional classes. The study suggests that metabolic networks have a hierarchical modularity that reconciles their scale-free topology, high clustering coefficient, and power-law scaling of C(k). This hierarchical organization is consistent with the idea that evolution acts at multiple levels, affecting both small and large modules. The findings suggest that hierarchical modularity is a general feature of biological networks, applicable to other cellular and complex systems.