IntroductionHIV self‐testing (HIVST) provides couples and individuals with a discreet, convenient and empowering testing option. As with all HIV testing, potential harms must be anticipated and mitigated to optimize individual and public health benefits. Here, we describe social harms (SHs) reported during HIVST implementation in Malawi, and propose a framework for grading and responding to harms, according to their severity.MethodsWe report findings from six HIVST implementation studies in Malawi (2011 to 2017) that included substudies investigating SH reports. Qualitative methods included focus group discussions, in‐depth interviews and critical incident interviews. Earlier studies used intensive quantitative methods (post‐test questionnaires for intimate partner violence, household surveys, investigation of all deaths in HIVST communities). Later studies used post‐marketing reporting with/without community engagement. Pharmacovigilance methodology (whereby potentially life‐threatening/changing events are defined as “serious”) was used to grade SH severity, assuming more complete passive reporting for serious events.ResultsDuring distribution of 175,683 HIVST kits, predominantly under passive SH reporting, 25 serious SHs were reported from 19 (0.011%) self‐testers, including 15 partners in eight couples with newly identified HIV discordancy, and one perinatally infected adolescent. There were no deaths or suicides. Marriage break‐up was the most commonly reported serious SH (sixteen individuals; eight couples), particularly among serodiscordant couples. Among new concordant HIV‐positive couples, blame and frustration was common but rarely (one episode) led to serious SHs. Among concordant HIV‐negative couples, increased trust and stronger relationships were reported. Coercion to test or disclose was generally considered “well‐intentioned” within established couples. Women felt empowered and were assertive when offering HIVST test kits to their partners. Some women who persuaded their partner to test, however, did report SHs, including verbal or physical abuse and economic hardship.ConclusionsAfter more than six years of large‐scale HIVST implementation and in‐depth investigation of SHs in Malawi, we identified approximately one serious reported SH per 10,000 HIVST kits distributed, predominantly break‐up of married serodiscordant couples. Both “active” and “passive” reporting systems identified serious SH events, although with more complete capture by “active” systems. As HIVST is scaled‐up, efforts to support and further optimize community‐led SH monitoring should be prioritized alongside HIVST distribution.