Seed dispersal is a critical life history stage of plants, yet its causes and consequences in tropical forest ecosystems remain poorly understood. I used a novel genetic technique, genotyping maternal tissue attached to dispersed seeds, to measure dispersal distance in a large natural tree population on Barro Colorado Island (BCI), Panama. I developed 10 microsatellite markers for Jacaranda copaia and used these markers to genotype seeds caught within seed traps in the 50 ha Forest Dynamics Plot on BCI. I found that a mixture model consisting of two different dispersal distributions, a Gaussian distribution describing local dispersal and an exponential distribution describing dispersal at longer distances, best-fit the observed distribution of seeds across two years in this species. Genetically measured distances showed that > 20% of seeds in each year moved distances > 100 m. Dispersal kernels differed between the years presumably due to differences in wind profiles between the years. Greater numbers of seeds were observed moving in one year due to greater seed production. Dispersal distances measured using the genetic markers were compared to estimates made using inverse models of seed shadows. I show that inverse modeling underpredicts the proportion of seeds moving distances > 100 m compared to the genetically determined dispersal distances. Finally, I examined the causes and patterns of individual variation in seed production and fecundity for three years in J. copaia. I found evidence in each year for a positive effect of individual canopy area and the fecundity of neighboring trees on seed production and differing effects of habitat associations on seed production. The study reveals that some species may be capable of moving a large fraction of their seeds to distances > 100 m. The implications of this finding are that sources of dispersal limitation in forest communities in some species may not solely be due to limited dispersal ability, but rather limited numbers of propagules, or variable numbers of propagules, available to be dispersed. The results further suggest caution in interpreting previous findings using inverse techniques, namely that as currently formulated, they may not adequately characterize the tail of the dispersal curve.