Seed germination index (GI) is a key indicator of plant toxicity and maturity for composting. A combined aerobic composting experiment was carried out in a selfdeveloped intelligent aerobic composting reactor system. The main materials were poultry manure digestate and pig slurry. The wheat straw and mushroom substrates were taken as bulking agents. Based on the obtained data of GI and the basic physicochemical parameters (volatile solid, VS; hemicellulose, HC; total carbon, CT; total nitrogen, NT; the ratio of total carbon to total nitrogen, CT/NT; lignin), Pearson correlation analysis and regression modeling were developed. The results showed that there were significant correlations （R≥0.83, Sig. was 0.000) between GI and the total volatile solid, total carbon, total nitrogen, hemicellulose and lignin contents on a dry basis, respectively. The unitary and binary linear models constructed had good degree of fitting （R≥0.81, Sig. was 0.000). The values of R and SEP of unitary linear models were (0.88, 9.75), (0.88, 10.32), (0.82, 12.73), (0.81, 12.77), (0.91, 8.23) and (0.91, 8.74) based on VS, HC, CT, NT, CT/NT and lignin, respectively. And the values of R and SEP of binary linear models were (0.92, 7.48) and (0.93, 7.58) using CT-NT and HClignin. In all calibrations, modeling using CT-NT as binary variables （R was 0.92, SEP was 7.58) had the best prediction efficiency. This study provides a methodology to support the rapid prediction analysis of GI. Although binary modeling using CT-NT had the best prediction efficiency, it was limited by the aerobic composting reactor volume and the number of samples obtained. Therefore, expanding the sample size should be needed to improve the model accuracy in the further research.