Two series of mouse chimaeras were produced by aggregating pairs of eight-cell embryos that differed at the Gpi-1s locus, encoding glucose phosphate isomerase (GPI-1); the paired embryos were respectively homozygous Gpi-1sa/Gpi-1sa and Gpi-1sb/Gpi-1sb. Chimaeric blastocysts were transferred to pseudopregnant females, that were homozygous Gpi-1se/Gpi-1se and produced only GPI-1 C enzyme. Quantitative electrophoresis of GPI-1 was used to estimate the contribution of each embryo (GPI-1 A and GPI-1B enzyme activity) to the foetus, placenta and other extraembryonic tissues of 12½ day chimaericconceptuses. For both series of chimaeras, the distributions of %GPI-1A in different tissues were classified as (1) balanced and typical, (2) balanced but atypical or (3) unbalanced. One series of chimaeras was clearly unbalanced, so that the cells derived from the (C57BL × CBA/Ca)F2 embryo (Gpi-1sb/Gpi-1sb) predominated over those derived from the BALB/c inbred strain (Gpi-1sa/Gpi-1sa) in most foetuses. Two significant observations were made concerning this unbalanced series. Firstly, the mean composition of the placenta and other extraembryonic tissues was similar to that in the foetus i.e. also unbalanced with (C57BL × CBA/Ca)F2 (abbreviated to BF2) cells predominating. Secondly, despite this generalizeddeficiency of BALB/c cells, there were differences in the frequency of non-chimaeric tissues between different developmental lineages. In 20/34 chimaeric conceptuses in the unbalanced series only BF2 cells were detected in the foetus, whereas both BF2 and BALB/c cells were present in at least one of the extraembryonic tissues. This group of chimaeras, therefore, shows some similarities to human confined mosaicism. Although chimaerism occurred more often in the primitive endoderm (hypoblast) lineage (yolk sac endoderm and parietal endoderm) than in the placenta, this may also be the case in human mosaics. The mosaic status of the human yolk sac endoderm is usually unknown so it is possible that mosaicism often occurs in the yolk sac endoderm as well as the trophectoderm in human ‘confined placental mosaicism’. The uniformly unbalanced phenotype seen in the mouse chimaeras may be a result of generalized cell selection against BALB/c cells in all tissues. As an alternative explanation, we propose that most of the BALB/c cells in the blastocyst are allocated to the mural trophectoderm, which has a limited mitotic potential and so contributes little to the mid-gestation conceptus. Further work is required to test these possibilities.