Completing a survey is often a novel experience for respondents. Instructions are placed to help respondents navigate through this task. Respondents however, donít always attend to the instructions, and in many cases ignore them completely (e.g., Gower and Dibbs, 1989), potentially increasing measurement error. Some theories have been developed to help guide researchers in questionnaire design (Jenkins and Dillman, 1995), but few specific recommendations exist on formatting instructions. The Teacher Listing Form (TLF) is the first in a sequence of ten questionnaires which comprise the Schools and Staffing Survey. The TLF contains a lengthy set of instructions detailing which teachers should be included and excluded from the form. In a previous round of cognitive interviews, Jenkins (1995) reported that some respondents read the instructions thoroughly, while other respondents skimmed them. In either case, respondents had to refer back to the instructions several times. This paper will present a modest qualitative effort to look at the implications of varying the instruction design. Specifically, by varying the instruction format and presenting respondents with either an attached set of instructions or a loose instruction card, we examined the influence the format had on respondentsí tendency to read the instructions in a cognitive interview.