Ant-dispersed plants usually produce seeds with appendages (elaiosomes), which are beneficial to ants. It is generally accepted that seed dispersal by ants (myrmecochory) is mediated by the presence of elaiosome that lures ants to transport the diaspores. Several hypotheses have related removal of seeds to nutritional needs of ants, and others have argued the importance of possible handles (i.e. structures that ants can grasp) in transporting by ants. To clarify the functional role of the elaiosome, ant behavior and seed removal rate were assessed by presenting simultaneously three types of items including entire seeds with elaiosome (SE), seeds without elaiosome (S), and seeds with an artificial "handle" glued to seeds (S+H) from five myrmecochorous species from four genera, i.e. Corydalis giraldii Fedde, Corydalis pseudoincisa C. Y. Wu, Chelidonium majus L., Viola grypoceras A. Gray, and Epimedium pubescens Maxim. to their two corresponding dispersing ants including Lasius alienus (Foerster) and Pheidole nietneri Emery in the nest surroundings in the field. Among the five studied plants, we found that the seed size of E. pubescens was significantly larger than that of Corydalis giraldii (P < 0.05). Both the seed weight and elaiosome weight of E. pubescens was the largest, and the elaiosome/seed mass ratio of C. giraldii was the highest. In addition, we found that no significant differences existed in the ant ‘Antennate’, ‘Examine’, and ‘Pick-up attempt’ behaviors among the three types of seeds from each plant species, excluding L. alienus, in which ‘Pick-up attempt’ of the S by ants was greater than that of both SE and S+H in Corydalis giraldii, Corydalis pseudoincisa, and Chelidonium majus plants. For the three types of seeds from each plant species, the removal rate of SE was significantly greater than that of both S+H and S in Corydalis giraldii, Corydalis pseudoincisa, Chenopodium majus, and V. grypoceras by ant L. alienus and in Corydalis giraldii, Corydalis pseudoincisa and V. grypoceras by ant Ph. nietneri. In addition, the removal rate of S+H was significantly greater than that of S in Corydalis giraldii by ant L. alienus. The results indicated that the presence of elaiosomes on seeds might influence ant behavior and their seed removal rate either by serving as an attractant for ants or by both acting as a handle and serving as an attractant for ants. The results further showed that the role of elaiosomes strongly varied depending on the species of each partner in ant-mediated seed dispersal process.