The past decade has seen tremendous growth in the clinical application of cell-based therapies, and the number of planned human clinical trials to evaluate these therapies continues to increase in number and scope at a rapid pace. A considerable effort on this front has been devoted to evaluating the therapeutic potential of mesenchymal stem cells (MSCs), which were initially characterized as connective tissue progenitors resident in bone marrow. MSCs are now known to possess potent tissue reparative properties that have been linked to secretion of paracrine-acting angiogenic, trophic, anti-inflammatory, and immunomodulatory factors. Accordingly, MSC-based therapies are being evaluated for the treatment of a broad array of ischemic, inflammatory, and immunological disorders. Nevertheless, knowledge regarding how the wide-ranging activities of MSCs vary between and are specified within populations remains largely unexplored. Lack of such knowledge makes it difficult to predict and/or control how sampling bias and ex vivo expansion of populations alters their biological activity and therapeutic potency. Herein, we discuss how heterogeneity of MSC populations may explain, in part, disparate outcomes in both experimental animal and human clinical trial data, and discuss several strategies to achieve more reproducible and efficacious outcomes for MSC-based therapies.