Macrophages within the tumor microenvironment (TME) exhibit a spectrum of pro-tumor and antitumor functions, yet it is unclear how the TME regulates this macrophage heterogeneity. Standard methods to measure macrophage heterogeneity require destructive processing, limiting spatiotemporal studies of function within the live, intact 3D TME. Here, we demonstrate twophoton autofluorescence imaging of NAD(P)H and FAD to non-destructively resolve spatiotemporal metabolic heterogeneity of individual macrophages within 3D microscale TME models. Fluorescence lifetimes and intensities of NAD(P)H and FAD were acquired at 24, 48, and 72 hours post-stimulation for mouse macrophages (RAW 264.7) stimulated with IFN-γ or IL-4 plus IL-13 in 2D culture, validating that autofluorescence measurements capture known metabolic phenotypes. To quantify metabolic dynamics of macrophages within the TME, mouse macrophages or human monocytes (RAW264.7 or THP-1) were cultured alone or with breast cancer cells (mouse PyVMT or primary human IDC) in 3D microfluidic platforms. Human monocytes and mouse macrophages in tumor co-cultures exhibited significantly different FAD mean lifetimes and greater migration than mono-cultures at 24, 48, and 72 hours post-seeding. In co-cultures with primary human cancer cells, actively-migrating monocyte-derived macrophages had greater redox ratios (NAD(P)H/FAD intensity) compared to passively-migrating monocytes at 24 and 48 hours post-seeding, reflecting metabolic heterogeneity in this subpopulation of monocytes. Genetic analyses further confirmed this metabolic heterogeneity. These results establish label-free autofluorescence imaging to quantify dynamic metabolism, polarization, and migration of macrophages at single-cell resolution within 3D microscale models. This combined culture and imaging system provides unique insights into spatiotemporal tumorimmune crosstalk within the 3D TME.