Standard procedures for conducting household-based surveys require obtaining a list of household residents. In self-administered census questionnaires, rosters have been used to convey information about who should be listed. However, residency rules do not necessarily follow respondents' intuitions about who should be considered a household member. The value of rosters rests, in part, on their effectiveness in conveying specific rules which the survey requires respondents to follow. The current paper reports on research designed to test the efficacy of five roster forms, containing different amounts and kinds of information, in conveying specific decennial Census rostering rules. Cognitive interviews (N=64) were conducted to determine how respondents completed a rostering task. To evaluate respondents' understanding of Census residency rules, 13 short vignettes, describing different residency situations, were employed. Respondents provided verbal explanations for each vignette response. Vignette responses are analyzed to evaluate respondents' conformity with Census residency rules. Content analysis of respondents' explanations is used to determine the extent to which information provided on rosters is utilized in making residency decisions. This paper demonstrates the usefulness of vignettes in assessing the effectiveness of various rostering techniques. It also discusses the utility of providing specific information as part of the rostering task.