In this chapter, we examine the effects of a complexity that is specific to self-administered surveys, the fact that respondents are often expected to answer certain items, but not others. Research has found that item non-response, the failure to answer items that should be answered, is greater in questionnaires that include branching instructions than in questionnaires which do not include them (Turner et al., 1992; Featherston and Moy, 1990; Messmer and Seymour, 1982). However, very little explanation for this is offered, except to say that items with branching instructions cause greater confusion. Consequently, we attempt in this chapter to offer a theoretical framework as to why branching instructions may be confusing, followed by an empirical test of some of the concepts. We end by interpreting the empirical results in light of a proposed model of the question-answer process applied to self-administered questionnaires.