This paper summarizes the research conducted by SRD between 2004 and 2008 to evaluate and improve the questionnaire planned for use on the handheld computer for the Nonresponse Followup (NRFU) for the 2010 U.S. Census. This paper provides an overview of results of three rounds of English cognitive testing, two rounds of Spanish cognitive testing, two rounds of behavior coding of successive versions of the NRFU instrument (both conducted in both languages), an observational study of the NRFU interview (also conducted in both languages), and two rounds of usability testing with both language instruments. Results from these different pretests led to the development of a revised NRFU questionnaire that was cognitively tested in English. This paper highlights key findings and recommendations from these different pretesting methods. The two largest modifications to the questionnaire based on testing were the creation of a topic-based instrument (as opposed to a person-based one) and the revision of the presentation of residence instructions from a flashcard to a series of questions. The revision to a topic-based sequence allows the interviewer to shorten questions for later people in the household. The need for this change was demonstrated through both cognitive testing and behavior coding. The revision of the residence instruction presentation is aimed at fixing the problem demonstrated through the observational study that interviewers do not often present respondents with the flashcard and through cognitive testing that when presented with it, respondents tend not to read the flashcard. Additionally, the question-answer sequence is expected (based on cognitive test findings) to produce better and more complete results, particularly with respondents with lowerlevels of literacy in both languages.