Digital addiction (DA) has been identified as an emerging public health problem worldwide. However, the extent and direction of the association between DA and interpersonal relationships (IRs) are unknown. Does DA have adverse effects on IRs, and how credible is the evidence for this association in published analyses of real-world data? Using the PRISMA method, we conducted a meta-analysis to quantitatively synthesize the results of the relevant studies and obtain reliable effect size estimates and performed an analysis of moderating factors. A systematic literature search identified 98 studies involving 134,593 participants and 99 effect sizes. A significant negative association was observed between DA and IRs. Importantly, our meta-analysis revealed that the DA subtype has no significant moderating effect on IRs, suggesting that combining numerous categories of DA rather than focusing on specific forms of DA may be appropriate for understanding the relationship between DA and IRs. Relative to the IR subtype, the association between DA and offline relationships is significant and negative, whereas the association between DA and online relationships is significant and positive. The strength of the relationship is also influenced by the participants' sex ratio, educational level, and measurement tools. These results may help resolve the disagreement over the magnitude and direction of the association between DA and IRs and have potential implications for the treatment of DA.