Self-reported disability prevalence often affects policy and resource allocation decisions and therefore requires reliable and valid measurement. This study evaluated 3 response formats on the reporting of disability for self and other household members. In a web survey, 60,873 U.S. respondents were randomly assigned to one of three groups: 1) presented disability items separately for self and household members using yes-no grid (YNG); 2) presented disability items separately for self and household using multiple response format (MRF); or 3) presented disability items and asked to select whether self, another household member, self and another household member, or no one in household had each impairment (Combination Grid - CG). Differences were obtained for the response formats in disability prevalence, self-reported severity of disability, and age of disability onset.