BackgroundAntenatal care (ANC) clinics serve as key gateways to screening and treatment interventions that improve pregnancy outcomes, and are especially important for HIV-infected women. By disaggregating data on access to ANC, we aimed to identify variation in ANC attendance by level of care and across vulnerable groups in inner-city Johannesburg, and document the impact of non-attendance on birth outcomes.MethodsThis record review of routine health service data involved manual extraction of 2 years of data from birth registers at a primary-, secondary- and tertiary-level facility within inner-city Johannesburg. Information was gathered on ANC attendance, HIV testing and status, pregnancy duration, delivery mode and birth outcomes. Women with an unknown attendance status were considered as not having attended clinic, but effects of this assumption were tested in sensitivity analyses. Multiple logistic regression was used to identify associations between ANC attendance and birth outcomes.ResultsOf 31,179 women who delivered, 88.7% (27,651) had attended ANC (95% CI = 88.3–89.0). Attendance was only 77% at primary care (5813/7543), compared to 89% at secondary (3661/4113) and 93% at tertiary level (18,177/19,523). Adolescents had lower ANC attendance than adults (85%, 1951/2295 versus 89%, 22,039/24,771). Only 37% of women not attending ANC had an HIV test (1308/3528), compared with 93% of ANC attenders (25,756/27,651). Caesarean section rates were considerably higher in women who had attended ANC (40%, 10,866/27,344) than non-attenders (13%, 422/3360). Compared to those who had attended ANC, non-attenders were 1.6 fold more likely to have a preterm delivery (95% CI adjusted odds ratio [aOR] = 1.4–1.8) and 1.4 fold more likely to have a stillbirth (aOR 95% CI = 1.1–1.9). Similar results were seen in analyses where missing data on ANC attendance was classified in different ways.ConclusionInner-city Johannesburg has an almost 5% lower ANC attendance rate than national levels. Attendance is particularly concerning in the primary care clinic that serves a predominantly migrant population. Adolescents had especially low rates, perhaps owing to stigma when seeking care. Interventions to raise ANC attendance, especially among adolescents, may help improve birth outcomes and HIV testing rates, bringing the country closer to achieving maternal and child health targets and eliminating HIV in children.Electronic supplementary materialThe online version of this article (doi:10.1186/s12889-017-4347-z) contains supplementary material, which is available to authorized users.