AbstractGhost fishing is the capacity of lost traps to continue to catch and kill animals. In the spiny lobster (Panulirus argus) fishery in Florida, the effects of ghost fishing are of particular concern, given the estimated 10s of 1000s of traps lost annually. We distributed 40 each of the three types of lobster traps (wire, wood–wire hybrid, and wood slat) at three locations in the Florida Keys to simulate ghost fishing. Divers monitored these traps biweekly for 1 year then monthly for two additional years, recording the time ghost traps remained intact and continued to fish, as well as the number of live and dead lobsters and other animals in each trap. Wood slat and hybrid traps remained intact and fished for 509 ± 97 (median ± median absolute deviation) and 480 ± 142 d, respectively. Wire traps fished significantly longer (779.5 ± 273.5 d, p < 0.001), and several fished until the end of the experiment (1071 d). Traps in Florida Bay fished longer (711.5 ± 51.5 d) than traps inshore (509 ± 94.5 d) and offshore (381 ± 171 days; p < 0.001) in the Atlantic Ocean. More lobsters were observed in hybrid traps (mean = 4.81 ± 0.03 s.e.) than in wood slat (3.85 ± 0.16) or wire traps (3.17 ± 0.03; F = 40.15, d.f. = 2, p < 0.001). Wire traps accounted for 83% of fish confined overall and 74% of the dead fish observed in traps. Ghost traps in Florida Bay and Atlantic inshore killed 6.8 ± 1.0 and 6.3 ± 0.88 lobsters per trap annually, while Atlantic offshore traps killed fewer (3.0 ± 0.69) lobsters, likely as a result of lower lobster abundance in traps. The combined effects of greater lobster mortality and greater abundance of lost traps in inshore areas account for the majority of the estimated 637 622 ± 74 367 (mean ± s.d.) lobsters that die in ghost traps annually.