Tetraploid (4n) embryo complementation assay has shown that embryonic stem (ES) cells alone are capable of supporting embryonic development (ES mouse), allowing the recovery of mouse lines directly from cultured ES cell lines. Although the advantages of this technique are well recognized, it remains inefficient for generating ES mice. In the present study, we investigated the effects of cell number of host 4n embryos on the production of ES mice. Four independent ES cell lines (two general ES cell lines and two nuclear transfer-derived ES cell lines) were used, and each cell line was aggregated with single (1×) to triple (3×) host 4n embryos. We found that birth rate of ES mice using 1× 4n embryos was quite low (0–2%) regardless of cell line, whereas except for one cell line, approximately 6–14% of embryos developed to full term in the case of 3× 4n embryos. Contamination of host 4n cells in ES mice was quite rare, being comparable to that generated using general methods even if they were delivered from 3× 4n host embryos. These results demonstrate that the use of 3× 4n embryos is effective for generating ES mice. Our technique described here will be applicable to any ES cell line, including general ES cell lines used for gene targeting.