In southern Africa, the commensal foraging cattle egret has a wide variety of potential ungulate species from which to select its hosts. We found that the preferred hosts involved just a few of the available species. We predicted that ultimately the selection of hosts should be adapted to maximize egret foraging success. Selection is hierarchical with birds first choosing a general area in which one or more host species may be present. They then choose a host herd, and finally select an individual which they accompany as long as its actions are suitable and prey items are disturbed. The decision process remains flexible. Egrets freely switch hosts moving from one cow or wildebeest to another. We found the predicted correlation between egret and host walking rate (steps/min.), and also that feeding success was related to number of egret steps. We looked for evidence that egrets would select species and individuals that allowed them to move at the optimum rate for foraging. We found egrets achieving maximum success at walking rates between 10 and 30 steps per minute. When hosts moved more slowly or more quickly, feeding success was low, and egrets switched hosts more frequently. We censused wild animals in game parks and domestic animals on farms, and found a similar overall abundance (individuals/km). Wild animal herds provided fewer suitable hosts (individuals moving at the optimal rate for the egrets). We conclude that there are three components to host selection. Hosts must be available in adequate numbers and must be easy to locate. They must be suitable in terms of providing egrets with good foraging situations. They must be reliable in terms of remaining in the same general area day after day and in terms of consistent rates of locomotion. Although wildebeest, zebras, and buffalo fit one or two of the criteria, cattle fulfill all three. We speculate that the widespread increase in cattle farming in the past century may have facilitated the world wide population explosion of the cattle egret.