The classic debate regarding the complex relationships between personal network, social media use, and mental well-being requires renewed examination in the novel context of pandemic-related social isolation. We present two surveys conducted at (i) the earlier months of the pandemic and (ii) the end of large scale social-lockdown measures in the U.S. to explore the social and behavioral antecedents of mental health states relating to social media use. Study 1 tracked the longitudinal changes of personal network, social media use, and anxiety level of a group of individuals (N = 147) over a three-month period during the pandemic. Study 2 replicated and extended the theoretical model to a race-representative U.S. adult sample (N = 258). Both studies consistently show that (1) more time on social media worsens anxiety. It also mediates the relationship between personal network size and anxiety. That is, a small personal network predicts more social media use, which is in turn related to increased anxiety. (2) Moreover, the effect of social media use on anxiety is mainly explained by news consumption on social media, rather than non-news related usage. (3) This link's strength is moderated by one's perception of COVID-19 impact, such that news consumption on social media increases anxiety more when the perceived impact is higher. These results demonstrate communication technologies' increasingly critical and multifaceted role in affecting mental health conditions.