During the 2010 Census, the U.S. Census Bureau will conduct several experiments to explore alternate content and ways to enhance data quality for the 2020 Census. One of these planned experiments is the Alternative Questionnaire Experiment (AQE), which tests alternate content regarding race and Hispanic origin for the 2010 Census mail questionnaire. This report documents the findings from two phases of cognitive pretesting that involved eight experimental race and ethnicity panels. The general research goal for the cognitive research was to examine respondent reactions to a number of new features. Key findings are that (1) respondents had no difficulty understanding and answering “combined” formats that collect data on Hispanic origin and race in a single question; (2) although most respondents found the examples helpful, White and Black respondents did not easily identify with a particular ethnicity; (3) respondents favored forms that contain the term “race” in the question stem rather than a more neutral, “Is this person... ” (4) respondents also favored spanners over the Asian and Pacific Islander categories. However, form navigation issues associated with these spanners were observed during the interviews; (5) most respondents considered the term “Negro” offensive.