The American Community Survey collects important housing, social, economic data from a sample of approximately three million households each year using a sequential multi-mode approach of mail, telephone, and personal visit interviewing. The U.S. Census Bureau conducted a field test in 2011 to assess the potential implementation of a fourth data collection mode: an internet response option. In preparation for that field test, in 2010 and 2011 the Census Bureau’s Human Factors and Usability Research Group conducted five rounds of pre-testing on the new Web version of the American Community Survey (ACS) and one round of testing on the Puerto Rico Community Survey (PRCS). We will discuss major findings, including participant usage of the progress indicator and duplication of household members on the roster, as well as differences discovered between the findings from the ACS and PRCS tests. We will also present quantitative data (e.g., completion time and satisfaction scores) and eye-tracking results from the testing and discuss suggestions for future research.