Little had been known about ETO protein until t(8;21) was found in 12%―15% of acute myeloid leukemia which resulted in AML1-ETO fusion protein. ETO protein has four conserved nervy homology regions termed NHR1― 4. A lot have already been known about NHR1, 2, 4: NHR1 is homologous with the Drosophila TATA-box-associated factor 110 (TAF110); NHR2 is a dimerization domain associated with mSin3A/HDAC; NHR4 is MYND class of zinc fingers associated with NCoR/SMRT/HDAC. Only the function of NHR3 remains unclear. In order to investigate whether NHR3 domain could participate in oligomerization, we cloned and purified this domain. Through gel filtration chromatography, dynamic light scattering and dissolved crystal electrophoresis, we found that NHR3 domain was a tight tetramer. Then we cloned NHR3+4 domain (i.e. NHR3 domain plus NHR4 domain), and discovered, by gel filtration chromatography and native PAGE, that NHR3+4 domain could form dimer in solution. This was the first time to ob- serve that NHR3 and NHR4 domains may have some con- tribution to the oligomerization of ETO protein, which might recruit corepressors in the form of dimer, and stabilize ETO dimerization through convergent strength of NHR2, NHR3 and NHR4 domains and then stabilize corepressors recruit- ment. These speculations are very worthy of further evalua- tion. Little had been known about ETO protein until t (8; 21) was found in 12% -15% of acute myeloid leukemia which resulted in AML1-ETO fusion protein. ETO protein has four conserved nervy homology regions termed NHR1- 4. A lot has already been known about NHR1, 2, 4: NHR1 is homologous with the Drosophila TATA-box-associated factor 110 (TAF110); NHR2 is a dimerization domain associated with mSin3A / HDAC; NHR4 is MYND class of zinc fingers associated with NCoR / SMRT / HDAC. Only the function of NHR3 remains unclear. In order to investigate whether NHR3 domain could participate in oligomerization, we cloned and purified this domain. Through gel filtration chromatography, dynamic light scattering and dissolved crystal electrophoresis, we found that NHR3 domain was a tight tetramer. Then we cloned NHR3 + 4 domain (ie NHR3 domain plus NHR4 domain), and discovered, by gel filtration chromatography and native PAGE, that NHR3 + 4 domain could form dimer in solution. This was the first time to ob- serv e that NHR3 and NHR4 domains may have some contribution to the oligomerization of ETO protein, which might recruit corepressors in the form of dimer, and stabilize ETO dimerization through convergent strength of NHR2, NHR3 and NHR4 domains and then stabilize corepressors recruit- ment These speculations are very worthy of further evalua- tion.