Lung cancers are a leading cause of death, with only a fraction of patients responding to treatment and a low survival rate. It is now well appreciated that the aberrant physical, chemical, and biological properties of the tumor microenvironment, particularly the immune landscape, are major contributors to heterogeneity in patient response. Conventional in vitro models have been successfully employed to understand migration of immune cells. However, many of these platforms lack vascular and stromal barriers, and are therefore unable to cohesively recapitulate features of the tumor microenvironment that are known to exacerbate immune evasion of solid tumors. While animal tumor models offer an opportunity to study immune cell trafficking and infiltration in a complex tumor microenvironment, they require high costs and logistical challenges of animal handling, while suffering from weak translatability to humans and ethical concerns. Broadly, high rates of attrition of clinical compounds during drug development have motivated a demand for disease-relevant, human cellular models that can be used in early discovery research. In response, we have developed microphysiological lung tumor models with vascular barriers (e.g. tubular or network structures) as preclinical tools for screening potential modulators of immune cell infiltration, specifically T-cells. I will provide examples of these models, discuss how we characterized and validated them, and contextualize the insight they offer us as part of a spectrum of in vitro models that vary in throughput and complexity. I will also highlight how complexity in these models can be leveraged to simultaneously probe efficacy and safety of potential therapeutics, enriching the information that can be obtained in early discovery efforts.