Social media represents an important and omnipresent social environment that impacts human health. However, little is known about when and how health-related social media use influences health behaviors. Based on three manuscripts, including a theoretical framework and four empirical studies, this dissertation comprehensively examined how specific types of health-related social media use might influence health behaviors, emphasizing effects on eating behavior. The research conducted in this dissertation is theoretically embedded in the reasoned action approach and self-determination theory. In the first manuscript, I outlined a conceptual framework to better understand mechanisms of action underlying health-related social media effects. According to this framework, social media use can be characterized by four essential factors: (1) communication features used, (2) directionality of interaction, (3) communicated contents, and (4) engagement level. These factors are essential for identifying the active ingredients of health-related social media use and communication that might cause health behavior change (i.e., the behavior change techniques that might be contained or whose enactment might be triggered). Behavior change techniques are, in turn, assumed to cause health behavior change through changes in psychosocial determinants of health behaviors. I also pointed out that social media contributes to increased social influences on health behaviors due to its omnipresence and high accessibility. It might also intensify these influences due to the unique characteristics of the social media environment (e.g., omnipresence, curation algorithms, network homophily). In the second manuscript, I experimentally examined the effects of posting about fruit and vegetable consumption on the intake of senders (posters) and receivers of these postings (Study 1, N = 81) and the senders’ intake in the context of a behavior change goal (Study 2, N = 128). I explored the effects’ underlying mechanisms of action (i.e., active behavior change techniques and their effect on psychosocial determinants of health behaviors) and the temporal dynamics of effects using intensive longitudinal data. Results show that public intake-related social media postings lead to higher fruit and vegetable intake of senders and receivers (Study 1). It further supported eating behavior change goals but not more strongly than private self-monitoring of intake (Study 2), suggesting that self-monitoring might be the most active behavior change technique. The examined psychosocial determinants of eating behavior did not mediate potential effects on eating behavior change. There were positive dose-response relationships within-person on a daily level. On days on which senders used social media more than usual related to their postings, they reported higher fruit and vegetable intake, perceived social support, and goal commitment (but lower self-efficacy). Furthermore, on days on which they posted more intake-related postings than usual, they reported higher fruit and vegetable intake (only in Study 2), intentions, self-efficacy, goal commitment, and more favorable attitudes. The results suggest that eating-related social media effects might be more transient and in timely proximity to actual social media use and need more time to unfold. The effects are complex (especially regarding the effects on psychosocial determinants) and likely dependent on the intensity of social media use and the social responses in reaction to postings. In the third manuscript, I investigated need-support provision in health-related social media communication as potential mechanisms of action for supporting health behavior change. I developed a short educational video about need-supportive communication strategies. The effects of watching the video on the use of these strategies in written responses to fictive social media postings were experimentally tested (Study 1, N = 76). I found that these strategies can be learned and applied in written communication immediately after watching the intervention video and one week later. The effects did not translate to a real- world setting, a forum-based health behavior intervention supported by an online support community (Study 2, N = 537). Participants who watched the intervention video did not show higher use of need-supportive communication strategies compared to participants who watched a control video. Due to this missing effect on strategy use, they did also not report higher perceived need-support, goal attainment (regarding fruit and vegetable intake and physical activity-related behavior change goals), and more favorable values in psychosocial determinants. There were positive effects on participant engagement (i.e., higher number of postings and subjective forum visit frequency in participants who watched the intervention video). Possible reasons for the missing effect on the use of the strategies might be the high goal attainment in both experimental groups, a misfit of the communication strategies and posting content, and low overall participant engagement. The solid theoretical foundation suggests the usefulness of need-supportive communication for supporting health behavior change. Thus, future research should identify populations that might benefit from the video intervention and increasing need-support (e.g., individuals with behavior change barriers) and foster the application of need-supportive communication strategies (e.g., through incentivizing role models or training of content moderators). This dissertation emphasizes the relevance of social media in influencing health behaviors and supporting health behavior change, highlighting the positive role of social media. When and how social media influences health behaviors is dependent on the actual enactment of effective motivation and behavior change strategies. Future research should take the complexity of social media effects into account and examine both, within- and between- person effects, and the interactivity between senders and receivers of social media communication. Therefore, researchers should make greater use of experimental methods with high ecological validity, intensive longitudinal data with longer time periods, big data approaches and combining different data sources, and advanced analysis methods such as social network analysis. Social media environments need to be considered to better understand influence factors for health behaviors and for effective health promotion and prevention.