Pancreatic cancer continues to have one of the poorest prognoses among all cancers. The drug discovery efforts for this disease have largely failed, with no significant improvement in survival outcomes for advanced pancreatic cancer patients over the past 20 years. Traditional in vitro cell culture techniques have been used extensively in both basic and early drug discovery; however, these systems offer poor models to assess emerging therapeutics. More predictive cell-based models, which better capture the cellular heterogeneity and complexities of solid pancreatic tumors, are urgently needed not only to improve drug discovery success but also to provide insight into the tumor biology. Pancreatic tumors are characterized by a unique micro-environment that is surrounded by a dense stroma. A complex network of interactions between extracellular matrix (ECM) components and the effects of cell-to-cell contacts may enhance survival pathways within in vivo tumors. This biological and physical complexity is lost in traditional cell monolayer models. To explore the predictive potential of a more complex cellular system, a three-dimensional (3D) micro-tumor assay was evaluated. Efficacy of six current chemotherapeutics was determined against a panel of primary and metastatic pancreatic tumor cell lines in a miniaturized ECM-based 3D cell culture system. Suitability for potential use in high-throughput screening applications was assessed, including ascertaining the effects that miniaturization and automation had on assay robustness. Cellular health was determined by utilizing an indirect population-based metabolic activity assay and a direct imaging-based cell viability assay.