Studies of seed dispersal and fruit production often use fruit traps. Different trap designs may give dissimilar estimates; however, prior to this study there has been no tropical forest field comparison of trap designs. Likewise, there are no recommendations about the number of traps required to assess ecological parameters, such as fruit production, mass and number of seeds dispersed, and number of plant species producing fruits. We compared the effectiveness of five trap designs in terms of fruit/seed bouncing out of traps, wind effects, area effects and seed removal by predators. These studies took place in Colombia in two tropical rain forests and in laboratory conditions. We found that 300 traps (0.085 m2 each) were not enough to obtain stable estimates in two out of four parameters (number of species and dispersed seeds). All estimates were highly variable when using fewer than 100 traps. All trap designs evaluated (mesh on PVC frame, hanging mesh, basin and funnel traps) prevent seed removal by predators, in sharp contrast with removal from the ground. Mesh traps were less affected by bouncing effects than plastic traps, and this factor was a large source of bias among estimates from different traps. Since up to 68% of dry mass may bounce out, it is important to consider adequate trap designs and to be careful when comparing studies using different methodologies. Small traps received fewer seeds per area, however area affects were not evident when bouncing effects were controlled for. We recommend the use of mesh traps on PVC frames, although hanging mesh traps are a good option in tropical forests without strong winds.