The purpose of this research was to investigate the criteria that people employ as they judge the relevance of bibliographic records and full-text documents. The dissertation developed a process model of relevance, which predicts change in people's reasoning structures as they progress from one stage of document evaluation to another. The dissertation also presents an a priori classification model that categorizes criteria along the dimensions of Topicality, Quality of Information, and Cognitive State. The research inquired into the use of criteria on two levels: 1) the micro-level, which followed the patterns of use for individual criteria, and 2) the macro-level, which pursued the patterns of use for classes of criteria. Multiple methods were employed in the research design. A laboratory experiment examined a predefined set of criteria, comparing participants' ratings on criteria importance after reading abstracts with those after reading full- texts. A naturalistic study mapped actual processes of document evaluation, concentrating on the frequency of use of criteria as people move from record to full-text evaluation in a natural time- space. The major results consisted of measurement of change in the use of both criteria and criteria classes across the two stages of document evaluation by means of importance ratings for the laboratory study and frequencies of use for the naturalistic study. The micro and macro level analysis identified the criteria and criteria classes that were rated as most important or used most frequently at both stages. The Hotelling T2 test on the laboratory data showed significant overall change in importance ratings of the three classes of criteria. Factor Analysis generated factor solutions that were informative for further development in criteria classification. The macro analysis for the naturalistic data extended the a priori model and established an eight class scheme for criteria. Participants' own reflections of their use of criteria were found to be mostly consistent with their actual use patterns. Finally, this research presents a taxonomy that classifies relevance criteria by their general nature and functionality. The findings of the research provide implications for advancing empirical research of relevance and relevance criteria and improving the design of bibliographic retrieval systems.