Objective To investigate the causal relationship between immune cells and various molecular subtypes of breast cancer (BC) by using two-sample Mendelian randomization (MR). Methods This study employed a two-sample MR analysis, with 731 immune cell types from the GWAS catalog as the exposure factors and four BC molecular subtypes (HR+/HER2−, HR+/HER2+, HR−/HER2+, and HR−/HER2−) from the BCAC Consortium as the outcome variables. For causal inference, inverse variance weighting (IVW) was employed as the primary analytical method, alongside multiple consistency validation approaches, including MR-Egger, weighted median, weighted mode, simple mode, and Bayesian weighted Mendelian randomization. Reverse MR and sensitivity analyses were employed to effectively evaluate and exclude potential biases and interference from reverse causality. Results Significant causal associations were identified for specific immune cell phenotypes across subtypes. In HR+ BC, the HER2+ subtype was associated with seven immunophenotypes (five protective and two risk increasing), whereas the HER2− subtype was associated with eight immunophenotypes (four protective and four risk increasing). In HR− BC, the HER2+ subtype was associated with five immunophenotypes (one protective and four risk increasing), and the HER2− subtype (triple-negative BC) was associated with seven immunophenotypes (six protective and one risk-increasing). CD28+CD45RA−CD8+ T cells and naive CD8+ T cells have a bidirectional causal relationship with HR+/HER2+ breast cancer, Whereas the expression of CD8 on CD39+CD8+ cells also has a bidirectional relationship with HR+/HER2− breast cancer. Furthermore, CD4−CD8− NKT cells exhibited opposing effect directions in HER2+ versus HER2− subtypes. Conclusion From a genetic perspective, this study confirms the existence of subtype-specific causal associations between immune cell phenotypes and distinct molecular subtypes of BC, revealing heterogeneity in immune regulation.