Substantial amounts of post-consumer scrap are exported to other regions or lost during recovery and remelting, and both export and losses pose a constraint to desires for having regionally closed material cycles. To quantify the challenges and trade-offs associated with closed-loop metal recycling, we looked at the material cycles from the perspective of a single material unit and trace a unit of material through several product life cycles. Focusing on steel, we used current process parameters, loss rates, and trade patterns of the steel cycle to study how steel that was originally contained in high quality applications such as machinery or vehicles with stringent purity requirements gets subsequently distributed across different regions and product groups such as building and construction with less stringent purity requirements. We applied MaTrace Global, a supply-driven multiregional model of steel flows coupled to a dynamic stock model of steel use. We found that, depending on region and product group, up to 95% of the steel consumed today will leave the use phase of that region until 2100, and that up to 50% can get lost in obsolete stocks, landfills, or slag piles until 2100. The high losses resulting from business-as-usual scrap recovery and recycling can be reduced, both by diverting postconsumer scrap into long-lived applications such as buildings and by improving the recovery rates in the waste management and remelting industries. Because the lifetimes of high-quality (cold-rolled) steel applications are shorter and remelting occurs more often than for buildings and infrastructure, we found and quantified a tradeoff between low losses and high-quality applications in the steel cycle. Furthermore, we found that with current trade patterns, reduced overall losses will lead to higher fractions of secondary steel being exported to other regions. Current loss rates, product lifetimes, and trade patterns impede the closure of the steel cycle. (C) 2016 Elsevier B.V. All rights reserved.