A study was conducted during 1996-2000 in an undisturbed tropical lowland rain forest site with an intact fauna at the Nouragues Biological Station, French Guiana, to demonstrate the potential of automated remote video monitoring for studying seed removal and seed fate. In addition, data in 1998 at Speulderbos, a mixed oak-beech forest in the Netherlands, were used. In Nouragues, a suite of different animals, including most of the ungulate and terrestrial rodent species occurring in the study area, and many other terrestrial animals. Acouchies (Myoprocta acouchy) were the first to discover seed plots in 96% of the cases and were always earlier than agoutis (Dasyprocta leporina). An important determinant of the ability of species to obtain seeds was their activity rhythm. Acouchies showed a strong apparent response to food abundance. Peccaries, in contrast, were only recorded around noon and only in low seed years, but this pattern may have been an artefact of the low number of encounters. Peccaries and squirrels removed few of the seeds (peccaries removed 5% of Carapa procera, squirrels 2% of Licania alba), yet they accounted for an important part of seed mortality. Acouchies accounted for majority of recorded removal in both Carapa (62%) and Licania (60%). Agoutis, finally, were less important seed removers than acouchies. Agoutis were two to six times heaviers than acouchies. No significant difference in dispersal distance between agoutis and acouchies was found at least not for the first 25 m. Both rapid removal and greater dispersal distance indicate a higher motivation of rodents to carefully store the seeds. If this pattern holds for other species and habitats, rapid seed removal could be a sign of higher-quality seed dispersal rather than higher predation. Agoutis and acouchies in French Guiana and wood mice in the Netherlands remove and disperse only seed at a time. The interval time between visits was proportional to the investment of rodents in scatterhoarding and their performance as seed dispersers, and can be used as a potential estimate of dispersal under incomplete seed retrieval.