The repeatability and precision for determining the metabolizable energy of duck feedstuffs (corn, soybean meal, cottonseed meal and wheat bran) using bionic digestive system were investigated for testing whether this system can be used for determination of nutrient digestibility. The single factor completely randomized design was adopted for the repeatability experiments in the same and among the three bionic digestive systems. In each repeatability experiment, 3 treatments (among 3 batches or 3 machines) with 10 replicates per feedstuff were used for determining the in vitro dry matter digestibility and metabolizable energy. The results showed that the within-batch, between-batch and total coefficient of variation of the dry matter digestibility and metabolizable energy of corn, soybean meal, cottonseed meal and wheat bran in the same bionic digestive system were all less than 1.40%, and the within-equipment, between-equipment and total coefficient of variation among these three bionic digestion systems were all less than 1.64%. The normality test and 95% confidence interval of dry matter digestibility and metabolizable energy in 4 duck feedstuffs showed that the maximum absolute bias of repeated measures were controlled in 0.94%, 1.00%, 1.02% and 0.88% in dry matter digestibility and 0.18, 0.29, 0.18 and 0.24 MJ/kg in metabolizable energy of corn, soybean meal, cottonseed meal and wheat bran, respectively. In repeated determination, the maximum absolute bias of 85% observed data of dry matter digestibility and metabolizable energy were not higher than 0.75% and 0.22 MJ/kg, respectively. According to the repeatability and precision of this experiment, a relatively high quantitative level for the determination of metabolizable energy of duck feedstuffs can be achieved using bionic digestion system.［Chinese Journal of Animal Nutrition, 2010，22（6）:1709-1716］