This paper presents an analysis of the theory and reality of sampling 'whole' families for a grounded theory research project exploring child and family management of acute childhood illness at home. Initial access to families for child health research projects can be difficult given the recent negative image of research involving children portrayed by the media. Once access to families is secured, whole family access may be prevented due to internal family dynamics in a society with rising rates of divorce and remarriage. Ethical issues also arise: individual family members may feel coerced into participating by other family members; if family members are interviewed separately researchers need to be particularly careful to maintain confidentiality; with children extra time is needed to facilitate rapport and an understanding of the researcher role. There is also a tension for children in the early stages of a research relationship between feeling safe and personal privacy. Theoretical sampling may direct data collection with wider members of the family, but it may not be possible to access these people. 'Whole' family research may be the aim, but inevitably each family will decide, collectively and individually, how this is determined.