Abstract.Zooplankton was sampled with emergence traps on aThalassiabed behind an insular reef at San Salvador. Bahamas. These traps tested the effect of sealing the trap to the substrate or allowing it to rest unsealed on the substrate. Total numbers of plankters collected did not differ between the two types of trap, but significantly more cyclopoid copepods were collected in unsealed traps and significantly more tanaids were found in sealed traps. Total densities in unsealed traps were inflated as much as 24–28%by contamination with holoplankton. Both trap designs, but particularly the sealed traps, collected animals that apparently crawled up the insides. A substantial number of animals, as many as 9–31% of those collected in sealed traps, may not be actually planktonic. Our results suggest that reefs produce fewer meroplankters per unit of substrate, and thus meroplankters contribute less to reef foodwebs than formerly thou