By reducing viral loads to very low levels for many years, ART offers the promise of being a useful tool in the prevention of HIV transmission. Success may require diagnosing and treating people as early as possible in the course of their infection. Some have argued for radical schemes involving universal testing and treatment to eliminate HIV altogether. There is a need to explore the level of evidence supporting such an approach. Key parameters have been determined by studying transmission in couples, and in epidemics. Studies of HIV transmission in couples have and continue to shed light on the biology of transmission. Some recent studies of concentrated European epidemics and of generalized African epidemics have been used to quantify epidemiological parameters, and highlight different types of failures in public health. Mathematical model-based approaches synthesize our understanding of transmission, and generally support the use of ART to mitigate the HIV epidemic. These predictions depend on some assumptions shown to be plausible, but have not yet been fully validated from empirical field data. These critically determine the success, or failure, of mass ART used to control HIV.