Survey respondents are sometimes asked to complete multiple questionnaires for multiple instances that are not enumerable by the researcher when a sample is designed. That is, the reporting units (individual respondents) are pre-identifiable and can be sampled, but the number and identities of the units of analysis (household members, subsidiary business entities, etc.) are only discovered during the survey itself. This condition presents questionnaire design challenges – the existing literature concerning one such design choice is reviewed: the topic-based approach (one question presented to the respondent for every unit of analysis simultaneously, often resulting in a grid format) versus a serial, unit-based approach (all questions presented to the respondent for one unit of analysis at a time). One extreme version of the serial approach is asking the respondent to complete multiple copies of a standalone questionnaire for each unit of analysis. This paper examines the behavior of respondents faced with such requests. Using paradata from GAO web surveys, the navigational patterns of within- and across-questionnaire activity are analyzed. Navigational patterns are varied and complex, and our research suggests that in some multiple-questionnaire surveys, the level of respondent effort may be very high – a substantial number of respondents approach such tasks by jumping back and forth between copies of instruments on a question-by-question basis, for example. We conclude with implications for survey burden and quality.