The authors consider the problems and possibilities presented by using unstructured observation in the home setting. The findings of the original study are described elsewhere (Briggs, Askham, Norman, & Redfern 1998; National Health Service [NHS] Executive 1998). In this article, the authors discuss process issues (e.g., gaining and maintaining access to the research setting, disengaging, what can and cannot be observed, and the validity of the data and analysis), drawing on a study of the nature of care as a form of social organization in the homes of people with dementia. They show that observation is not only possible but, when combined with conversational interviewing, essential for understanding the processes of caring and what it means to live with dementia.