Background and Aims Heterostyly and related style polymorphisms are suitable model systems to evaluate the importance of functional pollinators in the maintenance of population variability. In Narcissus papyraceus different functional pollinators, incompatibility system and flower morphology have been proposed to influence the maintenance of polymorphism through their effect on disassortative mating. Here a test is done to find out if the visitation rate of long- versus short-tongued pollinators correlates with the morph ratio and if the latter is related to other flower traits of the species across its main geographic range. Methods Floral traits from 34 populations in the south-west of the Iberian Peninsula and in north-west Africa were measured, perianth variation was described and a comparison was made of allometric relationships between sex organs and floral tube. Correlations between pollinator guilds, stigma–anther separation of reciprocal morphs (our proxy for disassortative mating) and morph-ratio variation were analysed. Finally, the incompatibility system of the species in the northern and southern borders of its distribution are described. Key Results Flowers from southern populations were significantly larger than flowers from centre and northern populations. The abundance of short-styled plants decreased gradually with increasing distance from the core region (the Strait of Gibraltar), with these disappearing only in the northern range. Although there was a significant difference in stigma–anther separation among populations, morph ratio was not associated with reciprocity or floral tube length. Long-style morph frequency increased with short-tongued pollinator visitation rate. Populations from both edges of the distribution range were self-incompatible and within- and between-morph compatible. Conclusions The style morph ratio changed gradually, whereas perianth trait variation showed abrupt changes with two morphotypes across the range. The positive relationship between the visitation rate of short-tongued pollinators and the decrease of the short-style morph supports our initial hypothesis. The results highlight the importance of different pollinators in determining the presence of style polymorphism.