Nanoparticles have an extensive range of diagnostic and therapeutic applications in cancer treatment. However, their current clinical translation is slow, mainly due to the failure to develop preclinical evaluation techniques that can draw similar conclusions to clinical outcomes by adequately mimicking nanoparticle behavior in complicated tumor microenvironments (TMEs). Microfluidic methods offer significant advantages over conventional in vitro methods to resolve these challenges by recapitulating physiological cues of the TME such as the extracellular matrix, shear stress, interstitial flow, soluble factors, oxygen, and nutrient gradients. The methods are capable of de-coupling microenvironmental features, spatiotemporal controlling of experimental sequences, and high throughput readouts in situ. This progress report highlights the recent achievements of microfluidic models to reconstitute the physiological microenvironment, especially for nanomedical tools for cancer treatment.