HIV Prevention Trials Network studies are testing a number of new technologies for preventing HIV infections and reducing AIDS morbidity and mortality, but strengthening existing antiretroviral therapy (ART) programs may be among the most promising ways to generate greater health benefits using available resources. A cascade to care for HIV-positive patients has been described-HIV testing, retention in pre-ART care, treatment initiation, and sustained suppression on ART-and it has been noted that many patients are lost at each stage. We constructed a detailed representation by combining data from different sources about each stage. We found that, although currently available data were not sufficient to specify several key aspects, the traditional model of the cascade could not fully reconcile trends in HIV testing, linkage to care, retention in pre-ART care, and retention on ART with the large numbers of persons on ART and the large percentage of patients initiating treatment at late stages of infection. We hypothesize that supplementing the traditional linear cascade model with patient health-seeking behaviors that allow patients who are not in pre-ART care to be initiated on ART, is essential to fully characterizing the current functioning of ART programs. We have termed this additional channel to ART as the "side door." Understanding the relative roles of the different channels to care will be important to intervening effectively to improve the cascade to care, and we propose several new types of data that should be collected. With these insights, it may be possible to considerably strengthen the impact of ART programs.