Global electronification has driven an unprecedented surge in electronic and electrical waste (e-waste), with approximately 75 % of this e-waste informally managed, releasing hazardous chemicals. Traditional association analyses have limited ability to establish causation due to inherent methodological limitations. Accordingly, causal machine learning was employed in this study to investigate causal relationships between exposures to ten classes of e-waste pollutants (including bisphenols, polycyclic aromatic hydrocarbons, phthalates [PAEs], organophosphate flame retardants, nitrogenous flame retardants [NFRs], volatile organic compounds [VOCs], primary aromatic amines [PAAs], light metals [LMetals], transition metals, and heavy metals [HMetals]) and six health biomarkers (including neutrophil gelatinase-associated lipocalin [NGAL], o,o'-di-tyrosine [diY], malondialdehyde [MDA], 8-hydroxy-2'-deoxyguanosine, 8-oxo-7,8-dihydroguanosine, and 8-oxo-7,8-dihydroguanine). Approximately one-third (17/60) of the pollutant-biomarker associations passed all refutation tests, suggesting potential causality. PAAs displayed the highest potential causal strength (4.87, variance explained for the outcome) on NGAL, with other pollutant-NGAL associations being negligible (< 0.5); PAEs on diY (121.68), far exceeding others (< 10); and HMetals (14.39), LMetals (11.75), PAEs (10.77), and PAAs (10.58) on MDA. VOCs, NFRs, and PAAs were potentially causally associated with biomarkers of oxidative DNA and RNA damage. Notably, some pollutants exhibited threshold effects (e.g., PAAs for NGAL at 5.00 μg/g and 11.25 μg/g creatinine). Overall, our analytic framework offers a cost-effective blueprint to strengthen causal inferences in observational studies, thereby informing effective interventions.Copyright © 2025. Published by Elsevier B.V.