Current in vitro models employed to study cancer biology at the preclinical phase rely on hard, flat surfaces that lack the complex array of physical and biochemical signaling cues provided by the architecture, mechanical properties, and heterotypic cell interactions within the tumor microenvironment. As such, drug candidates that appear promising during in vitro testing frequently fail in preclinical animal testing or worse, in clinical trials with patients. Osteosarcoma is the most common primary tumor of the bone and is characterized by a high degree of inter- and intra-tumor heterogeneity. This heterogeneity and the absence of physiologically relevant preclinical models has meant that developing new treatments for osteosarcoma has been especially slow. In this thesis, techniques from tissue engineering—using combinations of cells, scaffolds, and biological factors to create tissue constructs—were employed to model the osteosarcoma tumor microenvironment. First, mechanically tunable scaffolds were developed by coaxially electrospinning poly(ε-caprolactone)/gelatin core-shell fibers and were used to investigate the effects of three-dimensional architecture and substrate stiffness on osteosarcoma cell phenotype and response to treatment. Nuclear localization of YAP and TAZ increased as the substrate stiffness decreased, and the three-dimensional environment in part downregulated the IGF-1R/mTOR signaling cascade and altered the efficacy of combination therapy with doxorubicin and agents targeting IGF-1R/mTOR compared to monolayer controls. Then, having demonstrated that the architecture and mechanical properties alter the phenotype of osteosarcoma cells, tumor-associated macrophages, the most prevalent infiltrating immune cells in the tumor microenvironment, were introduced to the engineered model. Co-culturing osteosarcoma cells with macrophages within the scaffolds for 24 hours resulted in the microenvironment becoming highly inflamed, as indicated by elevated levels of tumor necrosis factor alpha (TNFα) and interleukin (IL)-6, and the effect was most pronounced upon moderately stiff scaffolds. Increased levels of TNFα and IL-6 correlated with resistance to doxorubicin treatment, and inhibition of STAT3 diminished inflammation-related resistance but did not improve the efficacy of doxorubicin. This work highlights the advantages that tunable and physiologically relevant models can bring to the study of cancer biology and the development of novel therapeutics.