ObjectiveTo gather preliminary insights through formative research on social media usage, and experiences, attitudes and perceptions around COVID-19 and COVID-19 vaccination in three high-risk, underserved districts in Karachi, Pakistan.DesignCross-sectional mixed-method design.Participants392 adults (361 surveys and 30 in-depth interviews (IDI)) from districts South, East and Korangi in Karachi, Pakistan.Main outcome measuresSocial media usage and knowledge, perception and behaviour towards COVID-19 infection and vaccination.ResultsUsing social media was associated with an increased probability of getting vaccinated by 1.61 units. Most of the respondents (65%) reported using social media, mainly to watch videos and/or keep in touch with family/friends. 84.76% knew of COVID-19 while 88.37% knew about the COVID-19 vaccination, with 71.19% reported vaccine receipt; reasons to vaccinate included belief that vaccines protect from the virus, and vaccination being mandatory for work. However, only 56.7% of respondents believed they were at risk of disease. Of the 54 unvaccinated individuals, 27.78% did not vaccinate as they did not believe in COVID-19. Despite this, 78.38% of respondents scored high on vaccine confidence. In IDIs, most respondents knew about COVID-19 vaccines: ‘This vaccine will create immunity in your body. Therefore, I think we should get vaccinated’, and over half knew how COVID-19 spreads. Most considered COVID-19 a serious public health problem and thought it important that people get vaccinated. However, there was a low-risk perception of self as only a little over half felt that they were at risk of contracting COVID-19.ConclusionWith our conflicting results regarding COVID-19 vaccine confidence, that is, high vaccine coverage but low perception of risk to self, it is likely that vaccine coverage is more a result of mandates and coercion than true vaccine confidence. Our findings imply that interactive social media could be valuable in fostering provaccine sentiment.