Abstract: Abstract: Liquid bioenergy from energy plants is a promising sustainable energy source in China. But, due to a huge population, it is impossible to use many cultivated lands to expand the planting area of liquid bioenergy to safeguard food security and meet rapidly increasing energy demand. Planting the energy plants on marginal lands in China is an important approach to develop liquid bioenergy industry and solve the supply problem of raw materials. It can be also helpful to relieve the pressure of carbon emission reduction and protect the environment. However, the availability and the multi-suitability of marginal lands, especially at a regional scale, remain uncertain. Accurate evaluation of marginal land potential for energy plants growing is a serious part for rational planning and utilization of marginal lands. Existing land multi-suitability evaluation models are mostly linear models, which often cause overestimation or underestimation of single ecological factor. A nonlinear ecological multi-suitability evaluation model was developed in this study, which introduces expert knowledge and uses Gauss curve to match the optimum ecological niche of the marginal land based on the demand of ecological niche of the energy plant. Matching degree is used as fitness value to measure the multi-suitability and potential of the marginal land for planting energy plants. Taking Guangdong Province as an example, 4 species of liquid energy plants, namely Cassava, Vernicia fordii, Jatropha curcas L. and Pistacia chinensis, were chosen to evaluate the potential of liquid bioenergy. Using a supervised classification process, we identified the marginal lands from the Landsat TM/ ETM satellite images with a 30 m resolution that cover Guangdong. The multi-suitability of the marginal lands for planting the 4 species was evaluated using the ecological niche models (ENMs), which incorporate local temperature, precipitation, soil, terrain and traffic. Our modeling consisted of 2 steps, one at the individual ecological factor level, and the other at the habitat level that integrated the outputs from the individual factors. Three kinds of situations were considered, which were represented by 3 types of function curves, namely the S-shaped, Bell-shaped, and Z-shaped curve. The S-shaped curve characterizes the situation that large ecological factor values favor the crop. The Z-shaped curve indicates that small factor values favor the crop. The Bell-shaped curve defines a most optimal range in the factor value, and the optimality drops as the factor value deviates from this range. The results support the following conclusions: the total area of the marginal land in Guangdong is 106.35×104 hm2, and the shrub land, open forest land, and grassland account for 31.36%, 43.86% and 22.43%, respectively. The shoal/bottomland and unused land only account for very small proportion. With abundant heat and water resources, good soil condition and rich marginal land resource, Guangdong Province has high multi-suitability of planting energy plants and great potential of developing liquid bioenergy industry. The area of marginal land suitable for planting the 4 species is 62.15×104 hm2, which accounts for 58.44% of the total marginal land. Ecological niche fitness value for Jatropha curcas L. and Pistacia chinensis is the highest, and for these 2 species have similar ecological niche, there is a competition between them for development space; the second is Vernicia fordii; Cassava has the lowest niche fitness value, which is suitable for growing on the west terrace plain and east coastal hilly tableland in Guangdong Province.