Although often considered as evolutionary dead ends, selfing taxa may make an important contribution to plant evolution through hybridization with related outcrossing lineages. However, there is a shortage of studies examining the evolutionary dynamics of hybridization between outcrossing and selfing taxa. On the basis of differential pollinator attractiveness, production and competitive ability of pollen, as well as levels of inbreeding depression, we predict that the early products of hybridization between outcrossing and selfing lineages will be F1s and first-generation backcrosses sired mainly by the outcrossing lineage, together with selfed F2s containing a limited genetic contribution from the outcrosser. These predictions were tested using amplified fragment length polymorphism and chloroplast markers to analyze the composition of a recent hybrid swarm between predominantly outcrossing Geum rivale and predominantly selfing Geum urbanum. In line with predictions, the hybrid swarm comprised both parental species together with F1s and first-generation backcrosses to G. rivale alone. Chloroplast data suggested that G. rivale was the pollen parent for both observed hybrid classes. However, there was no evidence for F2 individuals, despite the fact that the F1 was fully self-compatible and able to auto-pollinate. The pollen fertility of F1s was only 30% lower than that of the parental taxa, and was fully restored in backcross hybrids. Predicting future evolution in the hybrid swarm will require an understanding of the mating patterns within and among the mix of parental, F1 and backcross genotypes that are currently present. However, these results support the hypothesis that introgression is likely to be asymmetrical from selfing to outcrossing lineages.