Endogenous alcohol production can increase the blood alcohol concentration (BAC) of drowning victims following submersion and confound epidemiological studies of the role of alcohol. This study seeks to determine how soon after a drowning death a victim's BAC is influenced by post-mortem alcohol production. The drop in mean lung weight that occurs over time in the water was hypothesized to serve as a proxy for the time course of decomposition, and thus provide an empirical measure to determine how soon after death to first suspect endogenous alcohol. The autopsy lung weights of 562 previously healthy males who drowned were compared across six submersion time groups (0-11.9, 12-23.9, 24-47.9, 48-95.9, 96-167.9 and >or=168 h) and two times of year (winter and non-winter). The hypothesis that a drop in lung weight is sensitive to the time course of decomposition was supported by (1). a statistically significant drop in mean lung weight that occurred 12-23.9 h post-submersion in the non-winter months, but not until 96-167.9 h in the colder winter months; and (2). a significant drop in lung weight was not observed in the group of cases with zero BAC. With a parallel finding that an increase in the proportion of cases with a positive BAC first occurred at the 12-23.9 h submersion group during the warmer non-winter months, we concluded that production of alcohol can occur in bodies recovered from the water as early as 12 h after death. Because excluding drownings with submersion durations greater than 12 h would exclude almost half of our cases from epidemiological studies of alcohol and drowning, additional evidence from the forensic literature was used to develop an adjustment procedure to account for endogenous alcohol production for submersion times of up to 1 week.