The transplantable MXT mouse mammary tumor has been a useful tool for studying endocrine mechanisms underlying mammary tumor growth. It is our experience, however, that this model is unstable during serial transplantation. This paper analyses this variability from the viewpoints of histology and estrogen receptor content and indicates that these parameters should always be checked before planning experimental work. It is advised that a more homogeneous material is needed and that this goal should be achieved by clonal selection before transplantation.