Industrial symbiosis networks, characterized by complex inter-industry resource exchanges, propagate localized disruptions across multiple sectors, thereby generating greater systemic uncertainty than single-industry systems. This inherent complexity undermines reliable energy conservation and emission reduction (ECER) planning in promoting industrial symbiosis and heightens the risk of ineffective policies. In this study, we develop a probabilistic framework integrating Latin hypercube sampling and Sobol sensitivity analysis to assess the impacts of 389 uncertain factors on China’s iron and steel–thermal power–cement symbiosis network from 2020 to 2030. Our findings indicate that: (1) Nationwide industrial symbiosis could achieve energy savings of 47.1–61.2 Mtce and significantly reduce emissions of SO2 (170.3–248.9 kt), NOx (166.7–280.3 kt), and PM (41.8–96.7 kt) by 2030, with required investments ranging from 92.8 to 122.1 billion USD. By balancing ambition and feasibility, we establish policy-actionable ECER targets for the nationwide industrial symbiosis network based on a 75% feasibility threshold across 100,000 uncertain scenarios. (2) Conservation supply curve analysis shows that over 95% of the symbiotic technologies exhibit negative energy conservation costs under uncertainty, yet their cost volatility spans several orders of magnitude. Prioritizing 27 high-potential, low-cost technologies can capture 76.9% of the network’s total ECER potential. (3) Technology penetration rates—rather than intrinsic efficiencies—account for 84.5% of the variance in ECER outcomes, underscoring the importance of policies that foster widespread adoption of symbiotic technologies over incremental technological efficiency improvements. This study redefines industrial symbiosis planning from static roadmaps into dynamic adjustment mechanisms, offering robust support for promoting industrial symbiosis under uncertainty.