Background: Antiretroviral prophylaxis may be a critical strategy to reduce periconception HIV transmission. Maximizing the benefit of periconception pharmacologic HIV risk-reduction requires an understanding of the links between pregnancy and adherence to this prevention strategy. Methods: We assessed study gel adherence among women with pregnancies compared to women without pregnancies enrolled in the CAPRISA 004 phase IIB trial of 1% vaginal tenofovir gel. Pregnancy was assessed with monthly urine tests. Adherence was measured monthly and defined as proportion of sex acts covered by two returned, used applicators based on pre- and post-coital dosing. High adherence was defined as a median adherence score of >80%, that is, more than 80% of sex acts were covered by two applications of study gel. A multivariate generalized estimating equations (GEE) model with a binomial distribution was used to assess covariates associated with high adherence (>80%) over time. Median adherence before and after pregnancy was compared using Wilcoxon signed rank test. Results: Among 868 women, 53 had at least 1 pregnancy (4.06 per 100 woman years, 95% CI: 3.04, 5.31). Women with pregnancies had lower median adherence compared to women without pregnancies (50% [IQR: 45-83] vs. 60% [IQR: 50-100], p = 0.02). Women with pregnancies also had a 48% lower odds of high adherence compared to women without pregnancies when adjusting for confounders (aOR 0.52, 95%CI: 0.41-0.66, p<0.0001). Among women with pregnancies, adherence before and after pregnancy was not different (50% [IQR: 46-83] vs. 55% [IQR: 20-100], p = 0.68). Conclusions: Women with pregnancies were less likely to have high adherence to study gel compared to women without pregnancies. Understanding these differences may inform findings from HIV prevention trials and future implementation of antiretroviral prophylaxis for at-risk women who choose to conceive. The protocol for the parent trial is registered on ClinicalTrials.gov, NCT00441298, http://www.clinicaltrials.gov/ct2/show/NCT00441298. © 2013 Matthews et al.