The approaches to enlarge the protected areas are deeply embedded in the conservation planning. In practice, however, even in some sites of top conservation priority, there exist problems of inefficient conservation for lack of funding, to say nothing of assisting all species under threat from the viewpoint of conservationists. Identifying priority sites for conservation and establishing networks of minimum priority sites (NOMPS) are helpful for promoting the transition from number and size oriented, to quality and effectiveness oriented practices of biological conservation, and for realizing the target of biodiversity conservation with the most benefits for the least costs. Based on heuristic algorithm and integer linear programming (ILP), we propose a refined method of heuristic integer linear programming (HILP) for quantitative identification of the NOMPS to protect rare and endangered plant species (REPS) in Guangdong Province, China. The results indicate that there are 19 priority sites which are essential for protecting all of the 107 REPS distributed in 83 sites in Guangdong. These should be the paramount targets of financing and management. Compared with the ILP, which uses minimum number of sites as the only constraint, HILP takes into consideration of the effect of species richness, and is thus more suitable for conservation practices though with a little more number of priority sites selected. It is suggested that ILP and HILP are both effective quantitative methods for identifying NOMPS and can yield important information for decision making, especially when economic factors are constraints for biological conservation.