The paper introduces InstructNav, a zero-shot generic instruction navigation system designed to enable robots to follow diverse language instructions in unexplored environments. The system addresses the challenge of handling various navigation tasks by proposing a new planning paradigm called Dynamic Chain-of-Navigation (DCoN) and Multi-sourced Value Maps. DCoN unifies different types of navigation instructions into a standard planning framework, while Multi-sourced Value Maps model key elements in instruction navigation, converting linguistic DCoN planning into actionable robot trajectories. InstructNav demonstrates robust performance in zero-shot tasks such as R2R-CE, Habitat ObjNav, and demand-driven navigation (DDN), outperforming existing methods. Real robot experiments in diverse indoor scenes further validate the system's effectiveness and adaptability. The project webpage is <https://sites.google.com/view/instructnav>.The paper introduces InstructNav, a zero-shot generic instruction navigation system designed to enable robots to follow diverse language instructions in unexplored environments. The system addresses the challenge of handling various navigation tasks by proposing a new planning paradigm called Dynamic Chain-of-Navigation (DCoN) and Multi-sourced Value Maps. DCoN unifies different types of navigation instructions into a standard planning framework, while Multi-sourced Value Maps model key elements in instruction navigation, converting linguistic DCoN planning into actionable robot trajectories. InstructNav demonstrates robust performance in zero-shot tasks such as R2R-CE, Habitat ObjNav, and demand-driven navigation (DDN), outperforming existing methods. Real robot experiments in diverse indoor scenes further validate the system's effectiveness and adaptability. The project webpage is <https://sites.google.com/view/instructnav>.