Collection of antiretroviral medication adherence data in the homes of HIV-infected people may have methodological advantages that can improve data quality. However, the feasibility of this approach has not been established. In addition, data on adherence, and its predictors, among HIV-infected women have been limited. Sixty-three HIV-positive women who were prescribed at least one antiretroviral drug in the last month were interviewed in their homes. A standard instrument was used to collect data on all antiretroviral medications prescribed and taken in the three days prior to the interview. Data were also collected on factors thought potentially to affect the ability to be adherent. The results of this study suggest that it is feasible to conduct home-based adherence research. Sixty-seven per cent reported taking all prescribed antiretroviral medication doses. One-third took a sub-optimal dose putting themselves at increased risk of treatment failure and the selection of resistant HIV strains. Unintentional reasons for missing doses were most commonly reported. An ability to describe the intended effect of antiretroviral therapy on HIV viral load was the best predictor of adherence. This finding is consistent with other research suggesting that adherence is associated with an understanding and belief in the effectiveness of antiretroviral therapy.