The utility of pollination syndromes in predicting pollinators has been controversial. Flowers of Guihaiothamnus acaulis are tubular and vivid in color, indicating that butterflies might be the dominant pollinators of this species, based on the theory of pollination syndromes. To test this prediction, observations on the floral biology, pollinator behaviors, and breeding system were carried out in two wild populations. The results showed that diurnal and protandrous flowers of G. acaulis could last 7-10 days, and this species was self-incompatible. Thus, the fruit set was pollinator-dependent. In addition, pollen-consuming hoverflies and halictid bees were identified as the major pollinators of G. acaulis; butterflies were recorded as visiting the flower only once. The expanded corolla throat, massive pollen per flower, and high floral longevity suggest that G. acaulis had experienced the process of pollinator shift. Our results indicated that the actual pollinating fauna of plants were determined by complex factors including floral syndromes, the availability of pollinators, and historical adaptation to habitat. Pollination syndromes should be used carefully to predict pollinators of a particular flowering plant species.