Experimental chimaeras have been used to estimate the number of tissue progenitors from the variation in the proportions of cells of each genotype that can be found in a given tissue. However, because of a failure to appreciate the importance of assessing the extent of agreement required between the predictions made by a specific mathematical model and the actual data, we believe that the statistical procedures were inappropriate and have led to unjustified conclusions. We illustrate our argument by reference to published data on the number of progenitor cells in the facial nerve nucleus of the mouse (K. Herrup, T. Diglio, A. Letsou, (1984). Dev. Biol. 103, 329-336).