At-sea mortality information is important for understanding the magnitude of threats to protected sea turtle species. When a sea turtle dies, it typically sinks, starts decomposing, and will eventually float to the surface if the carcass remains intact and enough internal gases accumulate. However, few data are available regarding the rate and duration of these processes to allow estimation of time since death once carcasses are recovered. Twenty-seven Kemp's ridley (Lepidochelys kempii) and 15 green (Chelonia mydas) sea turtle cold-stunned carcasses were placed in wire mesh, weighted cages at varying water depths (10–40 m) and temperatures (18.5°C–28.7°C) in the northern Gulf of Mexico from June 2018 to October 2019. Cameras and temperature-depth-orientation recorders were used to document decomposition progression and carcass buoyancy. Decomposition rate was measured using corrected accumulated degree hours and values of observed time-to-float were compared with predictions based on laboratory and field experiments in previous research. Overall, carcasses did not float when deployed in waters > 30 m when temperatures were < 22°C and carcasses tended to float sooner in ≤ 20 m depths, especially if bottom temperatures were > 24°C. Green sea turtle carcasses floated in a variety of environmental conditions, but onset of positive buoyancy was not very predictable. Buoyancy of Kemp's ridley sea turtle carcasses was inconsistent, but float times were fairly predictable. We did not identify the exact depth at which sea turtle carcasses cannot generate enough gases to float, but that depth is likely very close to 40 m. Carcasses that became buoyant in ≥ 30 m depths tended to float for < 24 hrs before sinking again and, therefore, it is unlikely that they have enough time to drift to shore. This information enhances our understanding of the likelihood of carcasses washing ashore and can be incorporated into carcass backtracking analyses to facilitate identification of mortality causes.