In preparation for the 2005 National Census Test, cognitive testing of proposed changes on the Decennial Census short form was conducted. The changes that were tested were designed to address four characteristics of the form, including: 1) Space saving features such as layout modifications and shorter instructions; 2) Revised roster questions, the inclusion of three household level questions and one individual level question to replace residence rules instructions; 3) Revised instructions for the Hispanic origin item and the race item; and 4) revised instructions for Person 1. A total of 14 cognitive interviews were conducted in the Washington, DC metropolitan area to better understand the impact of these changes on how respondents interpret and respond to the revised questions. Our findings showed that of the space saving design features tested in the date of birth and age questions, the question on gender, and the question on relationship of household members to Person 1 did not have a negative effect on respondent behavior. We identified some problems with the three household-level roster questions and changes in the wording were recommended. Our findings also showed that the intent of the person-level coverage question (i.e., the question to identify where the person stayed on census day) was not clear to respondents. Therefore, we recommended that this question be replaced with a coverage question from another panel in the 2005 National Census Test. Respondents did not have difficulty with the revised instructions in the race question. Regarding the Hispanic origin question, we suggested minor formatting changes and further cognitive testing to ensure that the revised instruction is not offensive to persons of Hispanic origin. We also tested revised instructions to Person 1. Our findings indicated that modification in fonts and some changes in wording are needed. We recommended that these changes by subjected to further cognitive testing. Finally, our findings also indicated that the "continue note" instructions at the bottom of the first page should be repeated for persons 2-6. We recommended this change, if there is room on the form.