This article reviews research on the self-structure especially focusing on how positive and negative self-knowledge (elements) organize across multiple self-aspects. The view points of two related theories on this topic are fully reviewed and discussed: Compartmentalization of Self-Knowledge (Showers, 1992; Showers & Zeigler-Hill, 2012) and Intrinsic Dynamics of Self-Organization (Nowak, Vallacher, Tesser, & Borkowski, 2000). Both theories present similar views on how knowledge categories would emerge from the organization of positive and negative self-knowledge. However, due to the different research approaches, these two theories focus on the different aspects of this phenomenon. The theory of compartmentalization employs a card-sorting task, which contains forty adjective cards (20 positive and 20 negative). In this task, participants are asked to describe the different aspects of the self by sorting those cards into categories. The self-structure in this task is measured through the degree of compartmentalization. According to this theory, the self-structure can be largely classified as compartmentalized self-structure and integrative self-structure depending on its degree of compartmentalization. Compartmentalization of self-knowledge is regarded as relating to the stability of affective states, and can be seen as the strategies that people are used to cope with negative experiences. The theory of intrinsic dynamics of self-organization employs computer algorithm to simulate how positive and negative elements would interact with each other and eventually organize into different categories after several trials of interaction. Some elements are more influential than others, thereby categories are more likely to emerge according to the valence of elements with stronger influence. This theory also proposes that the press of integration is the underlying factor that drives the organization of positive and negative self-knowledge. In the end, this article briefly reviewed the research on the knowledge structure of others (e.g., close partners and parents). Although the theories regarding other-structures are largely consistent with the research on self-structures, some seemingly incompatible evidence also appeared. And this is exactly the direction that call for future research to clarify. Several questions related to this topic are discussed accordingly. (1) what is the mechanism that underlies the organization of positive and negative self-knowledge? The theory of compartmentalization demonstrated that the compartmentalized self-structure is highly related to emotional response categorization in which knowledge about the self is categorized based on their emotional responses (e.g., sad knowledge category, happy knowledge category), whereas the integrative self-structure is related to relatively unemotional semantic categorization in which knowledge is categorized based on their semantic meaning. However, there is evidence that can show emotional response categorization causes the compartmentalized self-structure, or vice versa. It is possible these two phenomena only occur concurrently. The theory of intrinsic dynamics proposes that it is the press for integration that drives the organization of positive and negative self-knowledge. However, this conclusion is reached through computer simulation, and the shortcoming is that it lacks supporting empirical evidence. (2) whether or not self-structures are the bases of affective disorders? Psychologists in clinical psychology often regard any mental structures as cognitive biases that underlie emotional disorders as depression and anxiety. However, psychologists in the personality and social psychology have demonstrated that self-structures are the strategies with which people are used to cope negative experiences. The present review takes the position of the latter and further proposes that self-structures and affective disorders are concurrently occurred phenomena. Therefore, the clinical research of affective disorders should not only consider environmental factors that cause or trigger the disorders, but also should try to counterbalance the degree of cognitive biases related to self-structures. In other words, both approaches may contribute to the alleviation or even cure of affective disorders.