The objective of the study was to evaluate the influence of pregnancy on the level of adherence with antiretroviral (ARV) drugs, in a prospective cohort of 72 pregnant women and 79 non-pregnant women. Adherence was measured by pill counting and self-reporting. Women were deemed adherent if 95% or more of all ARV had been taken as prescribed, in two occasions. According to pill counting, 43.1 and 17.7% of pregnant and non-pregnant women, respectively, met the criteria of adherence (P=0.001); in the postpartum, adherence declined to 20.6% (P=0.002). In both groups, adherence rates by self-reporting were significantly higher as compared with pill counting (P=0.001). In multivariate regression analysis, age > 29 years (odds ratio [OR] 3.58, confidence interval [CI] 95% 0.10-0.75, P=0.011), mean number of pills/day < 6 (OR 2.53, Cl 95% 1.07-6.01, P=0.035), and being pregnant (OR 3.33, Cl 95% 1.36-8.13, P=0.008) were independently associated to greater adherence.