Epiphyllous liverworts form a special group of bryophytes that primarily grow on the leaves of understory vascular plants in tropical and subtropical evergreen broadleaf forests. Being sensitive to moisture and temperature changes, epiphyllous liverworts are often consider to be good indicators of climate change and forest degradation. However, they are a poorly collected and taxonomically complicated group, with an only partly identified distribution pattern. In this study, we built four models based on 24 environmental variables at four different spatial resolutions (i.e., 1 km, 5 km, 10 km, and 15 km) to predict the past distribution of epiphyllous liverworts in China, using Maxent model and 63 historical location records (i.e., presence-only data). Both area under the curve of the receiver operating characteristic (AUC) and true skill statistic (TSS) methods are used to assess the model performance. Results showed that the model with the predictors at a 15-km resolution achieved the highest predictive accuracy (AUC=0.946; TSS=0.880), although there was no statistically significant difference between the four models (p > 0.05). The most significant environmental variables included aridity, annual precipitation, precipitation of wettest month, precipitation of wettest quarter, precipitation of warmest quarter, annual mean NDVI, and minimum NDVI. The predicted suitable areas for epiphyllous liverworts were mainly located in the south of Yangtze River and seldom exceeds 35° N, which were consistent with the museum and herbarium records, as well as the historical records in scientific literatures. Our study further demonstrated the value of historical data to ecological and evolutionary studies.