Spontaneously cycling LT/Sv strain female mice were mated to hemizygous Rb(X.2)2Ad males in order to facilitate the distinction of the paternal X chromosome, and the pregnant females were autopsied at about midday on the tenth day of gestation. Out of a total of 222 analysable embryos recovered, 165 (74·3%) were diploid and 57 (25·7%) were triploid. Of the triploids, 26 had an XXY and 31 an XXX sex chromosome constitution. Both embryonic and extra-embryonic tissue samples from the triploids were analysed cytogenetically by G-banding and by the Kanda technique to investigate their X-inactivation pattern. The yolk sac samples were separated enzymatically into their endodermally-derived and mesodermally-derived components, and these were similarly analysed, as were similar samples from a selection of control XmXp diploid embryos. In the case of the XmXmY digynic triploid embryos, a single darkly-staining Xm chromosome was observed in 485 (82·9%) out of 585, 304 (73·3%) out of 415, and 165 (44·7%) out of 369 metaphases from the embryonic, yolk sac mesodermally-derived and yolk sac endodermally-derived tissues, respectively. The absence of a darkly staining X-chromosome in the other metaphase spreads could either indicate that both X-chromosomes present were active, or that the Kanda technique had failed to differentially stain the inactive X-chromosome(s) present. In the case of the XmXmXp digynic triploid embryos, virtually all of the tissues analysed comprised two distinct cell lineages, namely those with two darkly-staining X-chromosomes, and those with a single darkly staining X-chromosome. Four X-inactivation patterns were consequently observed in this group, namely, (XmXp)Xm, (XmXm)Xp, (Xm)XmXp and XmXm(Xp) in which the inactive X is enclosed in parentheses. The incidence of these various classes varied among the tissues analysed. There was, however, a clear pattern of non-random selective paternal X-inactivation in yolk sac endodermally-derived samples which possessed two inactive X-chromosomes. This finding contrasts with the situation observed in the yolk sac mesodermally-derived and embryonic samples which possessed two inactive X-chromosomes, where the ratio of (XmXm)Xp:Xm(XmXp) was 1:1·20 and 1:1·03, respectively, being clear evidence that random X-inactivation had occurred in these tissues.