Recent human microbiome research has suggested that racial patterns between different groups of people can be understood as variation in how many and which microbes live in and on their bodies. Such racial classifications (from ‘Indigenous’ to ‘Black’ or ‘Caucasian’) are said to be helpful to better grasp microbiome-linked health-disparities (especially in the Global South) and diseases such as obesity and type 2 diabetes. In this paper, we argue that this assumption is illusive. We identify four different scenarios and argumentative patterns in current human microbiome research, which state that race matters for the field. However, we show that race should better be omitted in all these scenarios due to various conceptual and epistemic shortcomings. In addition, we acknowledge that there might still be an—admittedly minor—role for race to play in human microbiome research, namely in particular contexts and groups in which processes of racial self-identification matter for research participants. Based on this analysis, we urge scientists to reconsider the majority of racial classifications used in the field.