Perhaps the most fundamental question that faces the laboratory scientist is, “Which model system should I use to investigate the problem?” Failure to adequately address this issue can compromise even the most meticulous and inspired research program. If this is such a thorny issue, why use model systems at all? As with most biological systems, ovarian cancer is a complex disorder comprising tumor cells, stromal tissues, neovascularization, inflammatory responses, and other host responses to the tumor. Experimental science best progresses by controling all but a single variable and observing what occurs when that variable is modulated. Almost by definition, this requires a homogenous group of samples to work with. Using human cancer patients for research, it becomes rapidly apparent that the diverse nature of the tumors and the hosts greatly complicates such an approach, hence the development of various model systems. The problem with model systems is simple, they are models-not the true disease states, by their very nature they are less than perfect reflections of the way in which the system under investigation performs in vivo in the normal host. Model systems are essential research tools, but have to be used appropriately.