We have investigated age-related reactivation of the X chromosome by devising a model in which reactivation of a single gene in one cell among many can be identified. We have used mice with an X-autosomal translocation giving consistent non-random inactivation of the normal X (as judged by biochemical and cytogenetic techniques), that also carry a defective form of a histochemically demonstrable X-linked enzyme. When the gene for the normal enzyme was located on the inactivated normal X a uniformly negative histochemical picture would be predicted in doubly heterozygous animals. A very small proportion of enzyme-positive cells was found in young animals. This proportion increased very significantly with age, but the patch size did not change, showing that the result was not due to preferential division of enzyme-positive cells, but was instead due to the conversion of previously enzyme-negative to enzyme-positive cells. These observations provide the first evidence with a true X-linked gene for an age-related decrease in the stability of the X-inactivation mechanism.