Attempts to induce hosts and friends to prevent drinkers from reaching dangerous levels of alcohol impairment depend upon the ability to judge impairment from observable signs of physical appearance and behavior. In a study of this ability, researchers first observed and recorded signs of change in behavior and physical appearance among dosed drinkers in small social groups (n=149). Signs were grouped into impairment levels corresponding to three broad categories of blood alcohol concentration (BAC): <0.04% (no signs), 0.04%–0.08%, and >0.08%. Next, drinkers were then classified into judged impairment levels by guests observing small numbers in social groups (n=333), hosts observing large numbers in social groups (n=480), and servers observing patrons in public establishments (n=436). A random half of the observers in each setting were given instructional guidance in the relationships of signs to impairment level. Results showed all observers to exceed chance in their classifications of drinker impairment, with observations in the small social groups being significantly more accurate than those in the other two groups. A beneficial instructional guidance effect was significant for the social groups, with the greatest benefit found in detecting those over 0.04% in the small groups. The authors conclude that, while people are fairly accurate in judging alcohol impairment, their accuracy is greatest and guidance most effective in detecting the presence rather than degree of impairment and when attention can be concentrated upon a few drinkers at a time.