We have developed a technique of random primer extension of fixed chromosomes that is applicable to both mouse and man. Human chromosomes are not homogeneously labeled with this technique; those regions corresponding to R-bands appear to be more sensitive than those identified as G-bands, whereas centromeric regions are not labeled. These results not only corroborate specific structural differences between distinct regions of mammalian genomes but also open up the possibility of assays with specific primers to test whether primer extension is useful for the identification of genes and families of sequences on chromosomes.