The objective of this dissertation is to develop methods to draw relevant information from previously underexploited sources for behavioral inference in archaeology. The sources of information to be discussed are ceramics and formation processes. Ceramics have been the center of archaeological inquiry since the "Time-Space Revolution" during 1910's. Numerous studies have vigorously sought ceramics as a source of information for chronological, typological, and, more recently, locational inferences. In clear contrast, information encoded in ceramics about specific activities in the past has been surprisingly underexploited. This is because most extant ceramic analyses seldom have a perspective broad enough to recognize that those sherds are only fragments of once-functional tools. In this dissertation, extending the concept of tool kits, a method is proposed to treat a set of ceramics as tools to accomplish a certain task. Formation processes are another underexploited information source for behavioral inference. Initially, formation process theory was developed in reaction to studies by "new" archaeologists, who considered the archaeological record as a direct reflection of past human behavior. Owing to this historical reason, while this theory has demonstrated that formation processes must be an integral part of inferential processes, the role of information contained in formation processes tends to remain as negative, confounding factors. This dissertation proposes that information derived from formation processes can make more positive contributions to behavioral inference. Since formation processes, by way of the structure of refuse, encode qualitatively different aspects of past human behavior, an integration of such information with information about specific activities from once-functional artifacts would bring a fruitful result. An area of study that craves the exploitation of more information is small site analysis. Behavioral inference in small sites always suffers from the paucity of remains. Hampered by this limitation, conventional methods have failed to generate sufficient information for unequivocal behavioral inference at small sites. A specific analysis of Hohokam small sites is presented to demonstrate that the proposed methods are effective in exploiting relevant information from the same limited remains.