We test the hypothesis that there are host or environmental factors that significantly affect the likelihood of alcohol involvement in drownings. Our results are based on records of 234 drownings that meet predetermined eligibility criteria designed to exclude cases with postmortem blood alcohol concentrations (BACs) that do not reflect the BAC at the time of immersion. Cases are drawn from a total of 442 drownings occurring in Sacramento County, California, from 1974 to 1985. Overall, 41% of deaths were alcohol-associated; among these only one victim was under 15 years old. Among older persons, increasing age generally suggested a higher likelihood of alcohol involvement, and particularly of a BAC greater than 200 mg/dl. Other associated factors were male gender (OR = 2.5; 95% CI = 1.6, 3.8), activity (for land motor-vehicle occupants vs. all others, OR = 3.3; 95% CI = 2.6, 4.3), and time of year (January-June vs. July-December, OR = 2.1; 95% CI = 1.6, 2.8). A lower likelihood of alcohol involvement was seen for drownings in bathtubs (OR = .16; 95% CI = .04, .57) and swimming pools (OR = .47; 95% CI = .27, .82). Race was not a factor. Differing eligibility criteria have been used in studies of alcohol and drowning. After a critical review of the experimental literature, we propose that the following be adopted in future such studies: (i) death must occur within six hours of the onset of immersion, unless an antemortem sample is available and, unless evidence to the contrary exists, death can be assumed to have occurred within a few minutes of immersion; (ii) blood must be drawn for BAC determination within 24 hours of death.