Pollen flow is a key biological process that connects plant populations, preventing genetic impoverishment and inbreeding. Pollen-mediated long-distance dispersal (LDD) events are especially important for plant species in increasingly fragmented landscapes. Patterns of pollen dispersal were directly estimated and dispersal kernels modelled in an experimental population of Ranunculus bulbosus and Trifolium montanum to determine the potential for LDD. Eight and 11 microsatellite markers were used for R. bulbosus and T. montanum, respectively, to run a likelihood-based paternity analysis on randomly chosen offspring (Ntotal = 180 per species) from five maternal plants. High rates of selfing were found in R. bulbosus (average 45.7%), while no selfing was observed in T. montanum. The majority (60%) of mating events occurred at very short distances: the median of the observed dispersal distances was 0.8 m in both species, and the average distances were 15.9 and 10.3 m in R. bulbosus and T. montanum, respectively. Modelling the pollen dispersal kernel with four different distribution functions (exponential-power, geometric, 2Dt and Weibull) indicated that the best fit for both species was given by a Weibull function. Yet, the tail of the T. montanum pollen dispersal kernel was thinner than in R. bulbosus, suggesting that the probability for LDD is higher in the latter species. Even though the majority of pollen dispersal occurred across short distances, the detection of several mating events up to 362 m (R. bulbosus) and 324 m (T. montanum) suggests that pollen flow may be sufficient to ensure population connectivity in these herb species across fragmented grasslands in Swiss agricultural landscapes.