Tethering prey organisms continues to be a popular method of evaluating spatial and temporal variation of predator effects on population or community dynamics in marine environments. However, this technique relies on an untested fundamental assumption that there is a simple, reiiable relationship between losses of tethered prey and site-specific risk of mortaiity due to predation. To test this, we placed tethered and free grass shrirnp Palaemonetespugio together in mesocosms (1.3 m diameter) and varied the level of predation potential by exposing the prey to different densities of the murnmichog Fundulus heteroclitus, a common predatory fish in salt marshes of the eastern US. There were 10 replicate experimental runs, each involving 6 mesocosms that included 3 tethered and 27 free shrimp and either 0, 1, 2, 4, 8, or 16 predators. Survival of tethered shrirnp was assessed after 15, 30,45,60, 90, 120 and 180 min, but survival of free shrimp was determined only at the end of each run (180 min). In the absence of predators, average survival of tethered (93.3 %) and free (99.3 %) shrimp was similar. At the highest predator density, none of the tethered shrimp survived to 180 min, whiie suMval in the free group was 85.1 %. A significant interaction (ANOVA, F5,108 = 11.88, p < 0.0001) between tether treatment and predator density indicated a difference in the slopes of the relationships describing survival of free and tethered prey across the experimental gradient of predation potential. Chronological analysis of tethered prey losses across the experimental predation gradient also detected significant interactions (repeated-measures ANOVA, F", 324 = 3.24, p < 0.001) between the effects of predator density and exposure time. Tethering grass shrimp as a means of assessing even relative effects of predation by murnrnichogs on free prey is of questionable value because the relationship between relative loss rates of tethered and free s h r G p changed across a gradient of predation potential and varied with exposure time.