The nude mouse-human tumor model is presently considered to be a powerful tool in the development of new cancer therapies. This model has gained an increasing amount of interest since it was felt that murine tumor systems gave unsatisfactory results. 1 A major advantage is thought to be its capacity to select a new or to improve an existing therapy for a certain tumor type. 2 For this purpose, a panel of hum an tumor xenografts derived from a particular tumor type is being used so as to mimic a group of patients with that type of tumor. The optimism about the value of the model is based on the notion that treatment response obtained in panels of xenografts representing various tumor types is comparable to that in clinical practice. 2 , 3 However, taking into account that relevant characteristics such as cell cycle parameters, growth rate, metastatic spread, invasive properties, origin of stroma, pharmacokinetics, and metabolism in the tumor-bearing host are different from the situation in the cancer patient, 3 it is to be expected that some experimental results cannot be translated directly to man. Some investigators have indeed reported a lack of correlation between xenograft and patient response. 4 6 The aim of this chapter is to analyze the limitations of the nude mouse xenograft model for therapy studies. The deficiencies of the model should become the variables that define either more appropriate models or set the boundary conditions of the existing model.