The headwaters of the Clark Fork River in western Montana have been the focus of four decades of extensive and expensive remediation and restoration effort to reverse impacts of a century of mining, mineral transportation and processing. Historic mining spread dozens of contaminants via air and water, but federal and state agencies chose to focus cleanup on directly reducing exposures of a handful of the most acutely toxic heavy metals and arsenic. After three decades of positive response to remediation, trout populations have experienced precipitous, unexplained declines. Following convergent lines of evidence, I posit that bioaccumulation and biomagnification of selenium is the most likely explanation. 1) Early surveys showed Se present at elevated levels in smelter emissions, and other leachate sources are known, but monitoring of Se was largely suspended by the 1980s. 2) Fish declines are clearly associated with recruitment failure, and selenium toxicity is primarily manifest through embryo and larval mortality, also not directly monitored. 2) Literature review confirms that application of lime and phosphate fertilizer, treatments widely applied to immobilize heavy metals in soils, have the opposing effect of mobilizing Se. 3) Reducing heavy metals successfully facilitated the emergence of complex food webs in streams, which likely created the opportunity for bioaccumulation and biomagnification of Se, something not possible in past highly simplified food webs. 4) Recent riparian bird studies have substantiated nesting failure and elevated Se concentrations in body tissues. Because Se has not been consistently monitored in fish, soils, or water, direct evidence to confirm this hypothesis is lacking. The Se hypothesis impugns a prevailing narrative of restoration success, further contributing to the reluctance of managers to consider it. Divergent biogeochemistry and pathways of toxicity among multiple contaminants can fundamentally undermine or entirely negate restoration success in mining and industrially-contaminated ecosystems. Planning and adaptive management of restoration in mining and industry-impacted ecosystems must pay special heed to biomagnifying toxins like Se, mercury, and some persistent hydrocarbons, even if environmental concentrations appear relatively low, because restoration of food complex webs can set the stage for their toxic biomagnification.