Tetraploid (4N) complementation assay is regard as the most stringent characterization test for the pluripotency of embryonic stem (ES) cells. The technology can generate mice fully derived from the injected ES cell (ES-4N) with 4N placentas. However, it remains a very inefficient procedure owing to a lack of information on the optimal conditions for ES incorporation into the 4N embryos. In the present study, we injected ES cells from embryos of natural fertilization (fES) and somatic cell nuclear transfer (ntES) into 4N embryos at various stages of development to determine the optimal stage of ES cells integration by comparing the efficiency of full-term ES-4N mouse generation. Our results demonstrate that fES/ntES cells can be incorporated into 4N embryos at 2-cell, 4-cell and blastocyst stages and full-term mice can be generated. Interestingly, ntES cells injected into the 4-cell group resulted in the lowest efficiency (5.6%) compared to the 2-cell (13.8%, P > 0.05) and blastocyst (16.7%, P < 0.05) stages. Because 4N embryos start to form compacted morulae at the 4-cell stage, we investigated whether the lower efficiency at this stage was due to early compaction by injecting ntES cells into artificially de-compacted embryos treated with calcium free medium. Although the treatment changed the embryonic morphology, it did not increase the efficiency of ES-4N mice generation. Immunochemistry of the cytoskeleton displayed microtubule and microfilament polarization at the late 4-cell stage in 4N embryos, which suggests that de-compaction treatment cannot reverse the polarization process. Taken together, we show here that a wide developmental range of 4N embryos can be used for 4N complementation and embryo polarization and compaction may restrict incorporation of ES cells into 4N embryos.