Paranoia is a spectrum of fear-related experiences that spans diagnostic categories and is influenced by the environment. The extent to which traditional and digital media use is associated with paranoia remains unknown. We aimed to examine associations between media use and paranoia at the within- and between-person levels. Participants were 409 individuals diagnosed with schizophrenia spectrum or bipolar disorder. Measures included sociodemographic and clinical characteristics at baseline, followed by ecological momentary assessments (EMA) collected three times daily over 30 days. EMA evaluated paranoia and five types of media use: TV, social media, music, reading or writing, and other internet or computer use. Generalized linear mixed models were used to examine paranoia as a function of each type of media use and vice-versa. Characteristics associated with lower likelihood of media use included minority ethnoracial status, lower educational attainment, and diagnosis of schizophrenia vs. bipolar disorder. There were within-person associations between social media use and subsequently reduced paranoia in the schizophrenia group: fold-change=0.900; 95% CI: 0.840, 0.965; and between listening to music and subsequently decreased paranoia in the bipolar disorder group: fold-change=0.949; 95% CI: 0.906, 0.995. Associations of other media with paranoia were not significant. The associations of social media use and listening to music with reductions in paranoia may reflect clinical benefits of social interaction and distraction. However, systemic inequities and individual factors may hamper the accessibility of certain types of media.