The period of middle and high school is the key period of moral education, and gratitude is the key positive characters that moral education should focus on. Previous studies reveals that a high gratitude person are more positive emotions, mindfulness and high life satisfaction, et cetera. Numerous studies explored the mechanism of how gratitude influences subject well-being. In contrast, the studies of gratitude and psychological well-being are relatively less. Psychological well-being, which is different from subject well-being, developed from eudaimonism and focuses on self-fulfillment. Wood, Joseph and Maltby’s work(2009) indicated that psychological well-being is uniquely predicted by gratitude. However, how does gratitude affect psychological well-being? What mediators are there between them? The existing literature shows that social support and basic psychological needs are mediators between gratitude and subject well-being. Besides basic psychological needs plays an role of mediator between social support and psychological well-being. Thus, we presumed that social support and basic psychological needs might be the mediators between gratitude and psychological well-being. The purpose of this study was to test our hypothesis that social support and basic psychological needs mediated from gratitude to psychological well-being and further develop the mechanism of gratitude to well-being. 610 questionnaires were distributed and recruited, 583 of them were valid. In the valid questionnaires, the participants were 16.33(SD = .89) on average; 40.3% are male, 55.1% are female, and 4.6% did not report their gender; 37.0% were freshmen, 58.8% were sophomore, and the rest did not report it. They completed gratitude questionnaire GQ-6, perceived social support scale, basic psychological needs scale and psychological well-being scale. All these scales’ Cronbach α were acceptable: GQ-6 was .81, perceived social support scale was .86, basic psychological needs scale was .83, scale of psychological well-being was .94. Structural equation modeling with item-parceling was applied to analyze the data. The items of GQ-6 were parceled into 3 new indicators; the items of perceived social support scale and basic psychological needs scale were parceled into 3 new indicators respectively by using internal-consistency approach; the items of Ryff’s psychological well-being scale were parceled into 14 new indicators by using domain-representative approach. The results showed that: (1) The structural equation modeling revealed that the model fitted the data well (RMSEA=.071，CFI=.916，TLI=.905，SRMR=.055). Perceived social support and basic psychological needs mediated the effect of gratitude on psychological well-being. (2) In the chain-mediated model, the coefficient of the path from perceived social support to psychological well-being is non-significant. The chain mediating effect is about 33%. The direct effect from gratitude to psychological well-being is significant. It means there are some unrevealed mediators exist which we should further explore. Besides, the 6-dimension construct of Ryff’s psychological well-being scale is not supported by this study. The results of previous studies, especially the Chinese ones, about this scale’s construct validity are inconsistent. Therefore, we suggested researchers to revisit Ryff’s psychological well-being scale before they use it.