At present, most school gymnasiums are planned as a single-building with large-span structure. They are usually used by students for physical education classes and residents’ activities. Gymnasiums are also used by the residents as a refuge after the great earthquakes, such are of great importance. This research uses multivariate statistical analysis (Principal Component Analysis, Cluster Analysis, Coefficient of Correlation) and artificial intelligent theories (Support Vector Machine (SVM), Gene Expression Programming (GEP)) to explore the new seismic topics of school gymnasiums. The concept of principal component analysis was used to generalize the seismic factors by Eigen-values. Cluster analysis was adopted for gymnasiums’ clustering. Coefficient of correlation was utilized for the relationship between seismic factor and collapse ground acceleration. SVM was used to deduce the seismic performances of school gymnasiums. Finally, GEP was adopted for calculating the optimal seismic equation. The 479 school gymnasiums in Taiwan were used for research specimen. From the results know that the sequence of the relationship grades differs in every clustering. The RMSE values range from 0.0669 to 0.0856, the R2 values are between 0.7944 and 0.9183, show good results. The seismic equation was calculated by GEP, can provide for architects to design the gymnasiums at preliminary planning stage. The research methods also can provide academics for references.