Individual factors associated with HIV testing have been studied across multiple populations; however, testing is not just an individual-level phenomenon. This secondary analysis of 2005 and 2011 Ethiopia Demographic and Health Survey data was conducted to determine the extent to which the 2007 institution of an opt-out policy of HIV testing during antenatal care increased testing among women, and whether effects differed by women’s stigmatizing beliefs about HIV. A logit model with interaction between pre-/post-policy year and policy exposure (birth in the past year) was used to estimate the increased probability of past-year testing, which may be attributable to the policy. Results suggested the policy contributed to a nine-point increase in the probability of testing (95% CI 0.06–0.13, p < 0.0001). A three-way interaction was used to compare the effects of exposure to the policy among women holding higher and lower HIV stigmatizing beliefs. The increase in the probability of past-year testing was 16 percentage points greater among women with lower stigmatizing beliefs (95% CI 0.06–0.27, p = 0.002). Women with higher stigmatizing beliefs were less likely to report attending antenatal care (ANC), testing at their last ANC visit, or being offered a test at their last ANC visit. We encourage researchers and practitioners to explore interventions that operate at multiple levels of socio-ecological spheres of influence, addressing both stigma and structural barriers to testing, in order to achieve the greatest results in preventing HIV.