This study explores how communities in Zambia characterize vulnerable children in the context of HIV; demonstrates how estimates of vulnerability vary depending on definitions; and discusses the implications of these estimates for program delivery. Baseline research conducted in 2005 included cross-sectional community-based household surveys at six locations using multi-stage random sampling (totalling 1,503 households, reporting on 5,009 children) and participatory qualitative research (focus group and in-depth interviews) with adults and youth at four locations. Between 14 and 27% of children in the sample had experienced a parental death (2-5% maternal orphans, 7-13% paternal orphans, 4-10% double orphans). In addition, other characteristics that communities associated with children's vulnerability were prevalent: 26-34% had been taken into another household, 15-27% were living in female-headed households, and 11-28% were living in a household with someone who is chronically ill. Overall, 58-73% of children had one or more community-defined characteristics of vulnerability. This study highlights the need to carefully consider the meaning of "vulnerability" when targeting programmes to support children affected by HIV and AIDS. Local community input is vital to inform context-specific criteria for distributing programme resources. If used, eligibility criteria should be context-specific yet flexible to evolving community realities. In settings such as rural Zambia where levels of HIV-related vulnerability are high, it may be more efficient to target at the level of communities rather than assess individual households.