It is widely recognized that pharmaceutical marketing contributed to the ongoing US opioid epidemic, but less is understood about how the opioid industry used scientific evidence to generate product demand, shape opioid regulation, and change clinician behavior. In this qualitative study, we characterize select scientific articles used by industry to support safety and effectiveness claims and use a novel database, the Opioid Industry Documents Archive, to determine notable elements of industry and non-industry documents citing the scientific articles to advance each claim. We found that 15 scientific articles were collectively mentioned in 3666 documents supporting 5 common, inaccurate claims: opioids are effective for treatment of chronic, non-cancer pain; opioids are “rarely” addictive; “pseudo-addiction” is due to inadequate pain management; no opioid dose is too high; and screening tools can identify those at risk of developing addiction. The articles contributed to the eventual normalization of these claims by symbolically associating the claims with scientific evidence, building credibility, expanding and diversifying audiences and the parties asserting the claims, and obfuscating conflicts of interest. These findings have implications for regulators of industry products and corporate activity and can inform efforts to prevent similar public health crises.