To preserve biodiversity and achieve sustainable development is one of the main issues in ecology and highly important for our living environment. Brainea insignis (cycad-fern) is a precious and rare plant species, only occurring in the mountains of central Taiwan; therefore, it was chosen as a target for the study. The study attempted to predict the suitable habitat of cycad-fern in the Huisun study area by using multivariate statistics coupled with GIS techniques. The spatial distribution of the species was examined by overlaying the layer of cycad-fern samples collected by GPS on the layers of elevation, slope, aspect, terrain position, and vegetation index derived from SPOT images. Maximum entropy (Maxent) and decision tree (DT) models were developed and validated based upon three sampling designs with different combinations of cycad-fern samples taken from Sihwufongshan, Duhchuanling, and Kuandaushan trail in Huisun. Accuracy assessment indicated the three sampling designs had similar results, that is, the accuracies of DT were nearly equal to those of Maxent. They were highly efficient in implementation of model development and validation. Both DT and Maxent models greatly reduced the area of field survey to ferwer than 8％ of the entire study area at the first stage. Hence, DT and Maxent models were well suited for predicting the suitable habitat of the species. More importantly, these models have been validated once again with similar results by using independent samples from Kuandaushan trail in Huisun with a distance about 0.8 km from above-mentioned two sampling sites. Further confirmation will be done using independent samples taken from other sites far from the current sites in order to determine if the models are generally applicable at a larger spatial scale.