The concept of heterosis has already been put forward for a century. The hypothesis of Dominance, Superdominace and Epistasis has also been brought forward to explain the phenomenon of heterosis. As we know, there is spatio-temporal speciality about the expression of gene and only expressed genes contribute to the formation of heterosis. So the study on heterosis in expression level becomes more meaningful. A lot of studies on heterosis in this level have been done in plants, but there is no such study carried on animals in this area. In this study, the technique of mRNA Reverse Transcription Differential Display was used to research the heterosis molecular mechanism of animal. In order to expound the molecular genetic mechanism of animals heterosis, the 4 x 4 completely diallele cross experiment of 4 purebreds chicken was conducted among White Polymouth Rock (EE), Chinese Silk Chicken (CC), CAU Brown (DD) and White Leghorn (AA). The chicken of 16 cross combinations were reared to 8 weeks old, then 30 chicken in each combination were selected randomly and slaughtered. The traits of body weight of 8 weeks, wing weight, eviscerated weight, eviscerated weight with giblet, breast muscle yield, leg muscle yield, body length, abdomen fat weight, intramuscular fat width, tibia length were measured, and in which 8 individuals in each combination were selected randomly to collect the liver tissue samples, which were stored in liquid nitrogen or at -80 degrees C to be used for total RNA (TRNA) extracting. After the total RNA (TRNA) was extracted, 16 TRNA pools were formed in the same quantitative according to the concentration of 8 individual TRNA. They were reversely transcribed with three anchor primers H-T11 A, H-T11 G and H-T11 C. Then the reverse transcription PCR for each transcript product was done in two repeats at the same time with the same anchor primers and 8 random primers. The polyacrylamide gel electrophoresis of each PCR product was run in Bio-Rad Power 3,000 temperature control system. After electrophoresis, the gel was stained by AgNO3 according to the stain method described by Echt et al. The differential display bands in the polyacrylamide gel were counted. The band displayed is counted as 1 whereas no band is counted as 0 and only expression bands reproducible in two repeats were statistically analyzed. The correlation analysis between heterosis percentage and gene expression patterns was done with statistic analysis software (SAS) package. The statistic results indicated that among 690 total numbers of bands, the percentage of differential expression bands reproducible (457) is 66.23%. Eight kinds of gene differential expression patterns were found and listed as follow: 1): Band presents only in one purebred (P1);2): Band in one crossbred and its corresponding paternal purebred; or Band in one crossbred and its corresponding maternal purebred (P2);3): Band in purebreds and one crossbred (P3); 4): Band only in one crossbred (P4);5): Bands in both crossbreds and one purebred (P5);6): Bands only in both crossbreds (P6);7): Bands only in purebreds (P7);8): Bands both in purebreds and crossbreds P8. The differential expression of gene between purebred and crossbred chicken was detected for the first time. The proportion of each pattern in each kind of purebred combination is different. The percentage of P8 (75.34%) is the highest. The total percentage of differential expression patterns (24.66%) showed that the gene differential expression exists as a matter of fact. Among all the gene differential expression patterns, the percentage of P3 is the highest whereas the percentage of P7, P6 and P4 is very low, it indicated that different genes may have different expression patterns in purebreds and crossbreds. The results are similar to the study results on plants, which indicates that the gene differential expression between purebred and crossbred exists universally in biology. The correlation between gene expression patterns and heterosis percentage was studied, but correlation between P8 and the heterosis percentage is not significant (P > 0.05), it indicates that some patterns of gene differential expression may be the molecular genetic basic of heterosis. Among all the gene differential expression patterns, each pattern affects the expression of meat trait in different manner. There is significantly negative correlation between P4 and heterosis percentage of body weight of 8 weeks, breast muscle yield, leg muscle yield, eviscerated weight with giblet and eviscerated weight (P < 0.05); P1 is of significantly negative correlation with heterosis percentage of abdomen fat weight (P < 0.05) and of very significantly negative correlation with heterosis percentage of body length (P < 0.01); The negative correlation between P2 and heterosis percentage of intramuscular fat width is significant (P < 0.05); The positive correlation between P7 and heterosis percentage of leg muscle yield, wing weight, eviscerated weight with giblet and intramuscular fat width is significant (P < 0.05); The positive correlation between P5 and heterosis percentage of tibia length (P < 0.05) is significant. These results show that these 5 kinds of patterns play important role in heterosis forming of meat trait. P1 and P7 show that expressed gene in purebreds is depressed; P4 indicates that new gene expression occurs in crossbreds; P5 reveals that expressed gene only in one purebred express in all crossbreds. All genes of crossbreds come from purebred, which are not only the simple adding of these purebred genes, giving birth to unknown interaction between these genes coming from different purebreds, then leading to differential expression of genes. These gene differential expressions maybe form the heterosis of meat trait.