The recent alcohol involvement reduction in fatal crashes was studied using data from the Fatal Accident Reporting System (PARS). A new methodology was employed to estimate blood alcohol concentration (BAC) levels for all drivers and pedestrians in FARS where alcohol test results were unknown or unavailable. The percentage of drivers with BACs greater than or equal to .10% (legally intoxicated in most States) decreased from over 30% in 1982 to under 28% in 1984. This decrease was analyzed further to determine to whom and under what circumstances the alcohol decrease between 1982 and 1984 was the greatest. The findings are that the greatest decreases occurred in (1) multiple vehicle crashes, (2) daytime crashes, (3) weekday crashes and (4) crashes on high speed roadways; and among (5) teenaged drivers, (6) drivers who were restrained in the crash, (7) surviving drivers, (8) drivers with valid licenses, (9) drivers of intermediate and large size passenger cars, (10) drivers of 4-door passenger cars, (11) drivers of vans and special vehicles, and (12) drivers of 4-7 year old vehicles. The proportion of motorcycle drivers and adult pedestrians and bicyclists who were intoxicated (BAC greater than or equal to .10%) did not change appreciably between 1982 and 1984.Reasons for the alcohol decrease are discussed in terms of alcohol countermeasures and demographic and socio-economic factors. Host of the decreases suggest effects on responsible, social drinkers who have been influenced by the intense drunk driving publicity and increased enforcement procedures. Teenaged drivers and teenaged pedestrians were probably most affected by drinking age laws and grass roots citizen activist groups. Socio-economic factors appear to have contributed little to the decrease.