Background Accidental overdose is a major public health concern in North America with research primarily focused on cisgender men. Little is known about the burden of overdose among marginalised women, particularly in the context of child custody loss. This study aims to examine the prevalence of overdose and the association with child removal in a cohort of marginalised women. Methods This study draws on a merged dataset (2010–2018) of two community-based longitudinal cohorts of over 1000 marginalised women in Canada recruited using time-location sampling. After restricting to women who had ever had a live birth, bivariate and multivariable logistic regression using generalised estimating equations (GEE) were used to examine the association between child removal and overdose. Joint effects of child removal and Indigeneity were also investigated. Results Of the 696 women who reported ever having a live birth, 39.7% (n = 276) reported child removal at baseline. Unintended, non-fatal overdose rates were high, with 35.1% (n = 244) of women reporting ever having an overdose. Using bivariate GEE analyses, having a child apprehended and being Indigenous were positively correlated with overdose. Using multivariable GEE, child removal increased the odds of overdose by 55% (AOR: 1.55; 95% CI 1.01–2.39) after adjusting for education and Indigenous ancestry. Using multivariable joint-effects analysis, Indigenous women who had experienced child removal had over twice the odds of an unintended overdose than non-Indigenous women who had not lost custody after adjusting for education, food insecurity, and sex work (AOR: 2.09; 95% CI 1.15–3.79). Conclusion This analysis suggests that, after controlling for known confounders, women who have a child removed experience higher odds of overdose, and these odds are highest among Indigenous women. The high prevalence of overdose in this cohort suggests the need for further strategies to prevent overdose among pregnant and parenting women.