Increasing evidence shows that disseminated cancer cells (DCCs) can disseminate from early-evolved primary lesions much earlier than the classical metastasis models predicted and can enter a state of cellular dormancy at distant sites. The current paradigm suggests that the tissue microenvironment where DCCs lodge is a critical determinant of the timing of metastasis. However, how tissue-resident myeloid cells control this process is unclear. The lung tissue contains a resident population of homeostatic alveolar macrophages (AMs), whose function in metastatic dormancy and growth is poorly understood. Here, we reveal DCC heterogeneity and plasticity in the lung across disease evolution. We found a previously unrecognized role of mesenchymal- and pluripotencylike (M-like) programs in coordinating early cancer cell spread and a long-lived dormancy program in early DCCs. To test whether homeostatic AMs in the lung regulate the dormancy state of cancer cells in vivo, we depleted AMs in mouse models of breast cancer metastasis, using pharmacologic and genetic approaches. Following AM depletion, dormant DCCs began to proliferate and formed clusters and overt metastases. We further showed that in vitro AMs are responsible for inducing an M-like phenotype in cancer cells (of both early and late evolved cancer cells) and exclusively suppressed vi growth of early lesion MMTV-HER2 cells. Interestingly, AM function appears to be dependent on either the stage of the disease or the type of cancer. Our results suggest that AMs play a key role in limiting metastatic expansion of breast cancer cells via the induction of a DCC dormancy-like phenotype linked to mesenchymal programs. Our study allowed us to determine the specific contribution of the alveolar macrophage population to early DCC biology and metastatic dormancy.