In today’s fast-paced business environment, organizations rely more on employees’ voluntary participations and dedications. Following this trend, researchers have been investigated effective leaderships to transcend subordinates’ own interest for organization’s goal and achieve their performance beyond expectations for the past 30 years. Among these leadership behaviors, self-sacrificial leadership (SSL) is regarded as an inspirational and exemplary behavior. According to previous studies, SSL is linked to subordinates’ charisma perception of leader, organizational commitment, and leadership effectiveness. Although much is known about the SSL, there is still uncharted territory. Accordingly, this study had three objectives to explore this area: First, I sought to ascertain a relationship between SSL and organizational citizenship behaviors beneficial to the supervisor (OCBS). Second, I investigated whether the impact of SSL on OCBS is mediated by supervisor liking (SL). Third, I tested the negative moderating effect of subordinate’s learning goal orientation (LGO). Thus, we developed and empirically tested the following hypotheses: H1: SSL is positively related to OCBS. H2: SL mediates the relationship between SSL and OCBS. H3: LGO moderates the relationship between SL and OCBS. H4: LGO moderates the positive indirect effect of SSL on OCBS (through SL). Specifically, higher level of subordinate’s LGO attenuates the extent to which SL mediates the indirect effect of SSL on OCBS I used established, validated scales for all variables except SSL. I developed a 12-item scale of SSL reflecting Choi & Mai-Dalton (1998)’s conceptual definition of SSL. This 12-item measure has multi-item subscales corresponding to 3 dimensions: (1) division of labor, (2) distribution of rewards, and (3) exercise of power. Confirmatory factor analysis verified that 3-factor structure and 3 first-order factors plus one second-order factor had acceptable fit. This analysis represents that the dimensions are distinct, but also collectively reflective of the overall construct, SSL. Questionnaires were distributed to supervisor-subordinate dyads in Korean firms by postal mails or direct visits. All of the items except demography were measured on a seven-point Likert scale (ranging from 1 = strongly disagree to 7 = strongly agree). A total of 255 matched dyadic data was used for analysis. In the employee sample, 74% were men and the average age was 34.7 years (SD = 5.9). Around 51% of the respondents had at least a bachelor’s degree and their average tenure was 6.5 years (SD = 5.3). In the supervisor sample, 93% were men and the average age was 44.3 years (SD = 5.5). Approximately 53% of the respondents had at least a bachelor’s degree and their average tenure was 12.4 years (SD =7.3). Hierarchical regression analysis was used for testing hypotheses. Specifically, we adopted the three tests suggested by Baron and Kenny (1986) to explore the mediating roles of SL. We also used SPSS macro (MODMED) to verify the moderated mediation model. The result of analysis showed the positive relationship between SSL and OCBS, and exhibited that the effect of SSL on OCBS was completely mediated by SL. We also found that LGO moderated the relationship between SSL and OCBS. Finally, study results supported the hypothesized moderated mediation model, demonstrating that the magnitude of the indirect effect was contingent upon the level of subordinate’s LGO. This study has several implications. First of all, this is the first study to examine the relationship between SSL and OCBS, to my knowledge. This finding confirmed the effectiveness of SSL in organizational setting empirically. Second, I proved that subordinate’s affective reaction act as an important bridge between SSL and OCBS. That is, followers of leader who engaged in self-sacrifice like more his or her leader, resulting in higher OCBS. In other words, I proved one of the underlying mechanisms for explaining the effects of SSL on OCBS. Third, this study underlined the importance of LGO by showing the moderating role and moderated mediation effect. Of course, the present study also has its weakness. First, the results of this study are limited by its cross-sectional design. Second, various moderators and mediators are not considered in this paper. Third, I didn’t include group or organizational level constructs in research model. Finally, there is still a paucity of research on SSL’s antecedents which have potential influence on SSL. I hope that future researches are conducted to remedy these shortcomings and enlarge our understanding of dynamics associated with SSL.