Interdependent critical infrastructure systems represent substantial financial investments and are vital to maintain a basic level of social and economic well-being, making them attractive targets for malevolent actors. Many of these systems carry multiple products, each with unique needs and importance to different stakeholders. Tri-level optimization models have been proposed to capture the scale of a system’s resilience, representing the optimal actions taken by a defender to harden the system, by an attacker to interdict the system, and then by the defender to assign work crews for restoration, all under a limited budget. However, most prior work focuses on networks with a single product. This work extends a tri-level protection-interdiction-restoration model from a single commodity to multiple commodities, solving the model with a Benders’ decomposition and set covering decomposition. We propose a method to limit unmet weighted demand across commodities, taking into account unique interdependencies between network components and commodity-specific capacity requirements. An optimal solution is found iteratively by alternately fixing protection and interdiction variables. This work is illustrated with a case study of interdependent Swedish power and railway systems. Results demonstrate the convergent behavior of the master and subproblems, the value of network hardening, and the non-uniform network recovery trajectory. The proposed model is easily adapted to different commodity types, attack and defense budgets, crew availability, and commodity weights.