A cross-sectional survey was performed in 24 systems of care providing antiretroviral medications in Ethiopia, Kenya, Rwanda, Tanzania, and Uganda to examine current practices in monitoring rates of treatment adherence and defaulting. Only 20 of 48 facilities reported routinely measuring individual patient adherence levels; only 12 measured rates of adherence for the clinic population. The rules for determining which patients were included in the calculation of rates were unclear. Fourteen different definitions of treatment defaulting were in use. Facilities routinely gather potentially useful data, but the frequency of doing so varied widely. Individual and program treatment adherence and defaulting are not routinely monitored; when done, the operational definitions and methods varied widely, making comparisons across programs unreliable. There is a pressing need to determine which measures are the most feasible and reliable to collect, the most useful for clinical counseling, and most informative for program management.