The lymphomas probably represent the most complex and heterogenous set of malignancies known to cancer medicine. Underneath the single term lymphoma exist some of the fastest growing cancers known to science (i.e Burkitt's and lymphoblastic lymphoma), as well as some of the slowest growing (i.e. small lymphocytic lymphoma [SLL] and follicular lymphoma). It is this very biology that can dictate the selection of drugs and treatment approaches for managing these patients, strategies that can range from very aggressive combination chemotherapy administered in an intensive care unit (for example, patients with Burkitt's lymphoma), to watch and wait approaches that may go on for years in patients with SLL. This impressive spectrum of biology emerges from a relatively restricted number of molecular defects. The importance of these different molecular defects is of course greatly influenced by the intrinsic biology that defines the lymphocyte at its different stages of differentiation and maturation. It is precisely this molecular understanding that is beginning to form the basis for a new approach to thinking about lymphoma, and novel approaches to its management. Unfortunately, while our understanding of human lymphoma has blossomed, our ability to generate appropriate animal models reflective of this biology has not. Most preclinical models of these diseases still rely upon sub-cutaneous xenograft models of only the most aggressive lymphomas like Burkitt's lymphoma. While these models clearly serve an important role in understanding biology, and perhaps more importantly, in identifying promising new drugs for these diseases, they fall short in truly representing the broader, more heterogenous biology found in patients. Clearly, depending upon the questions being posed, or the types of drugs being studied, the best model to employ may vary from situation to situation. In this article, we will review the numerous complexities associated with various animal models of lymphoma, and will try to explore several alternative models which might serve as better in vivo.