Embryos produced by nuclear transfer from a patient's somatic cell offer one potential source of embryonic stem cells for treatment of human degenerative diseases. As with all of the approaches to such therapy, this has both strengths and weaknesses. The cells would be histocompatible with the patient's cells, be expected to have a normal life span, and in principle be a source of any other cell type. However, the time taken and the costs involved in the isolation of the appropriate cell population would probably prohibit large-scale application. Clones have been produced from the cells of adults of five species, but similar studies in at least five other species have produced early embryos, but not offspring. A variety of somatic cells have been used as successful nuclear donors. The present procedures have proved to be repeatable, but are very inefficient when typically between 1% and 4% of reconstructed embryos develop to adulthood. The inefficiency is the accumulated effect of failure at all stages of development. There may be differences between species and donor cell type in the precise pattern of loss. This outcome is assumed to reflect the inappropriate expression of a large number of genes whose lethal effect is exerted at different stages. Improvements in the efficiency may depend upon understanding those mechanisms in the early embryo that establish the precise chromatin structure that governs development.