THE STUDY EVALUATED BREATH ANALYSIS IN ALCOHOL-FREE SUBJECTS; THE COURSE OF BREATH ALCOHOL IN-VIVO IN RELATION TO ETHANOL METABOLISM; THE PRECISION OF THE ANALYSIS; THE CORRELATION BETWEEN BLOOD-ALCOHOL AND BREATH-ALCOHOL CONCENTRATION; AND THE CONFIDENCE LIMITS IN PREDICTING BLOOD ALCOHOL USING THE INTOXIMETER. THE SUBJECTS WERE 21 MALES WITH A MEAN AGE OF 42 YEARS, A MEAN WEIGHT OF 84 KILOGRAMS, AND ALL ACCUSTOMED TO MODERATE DRINKING. THEY WERE GIVEN STANDARD DOSES OF WHISKEY. THEN BLOOD AND BREATH SAMPLES WERE TAKEN AND ANALYZED USING AN AUTOMATED ENZYMATIC TECHNIQUE AND THE GAS CHROMATOGRAPH INTOXIMETER. THE GAS CHROMATOGRAPH INTOXIMETER WAS CALIBRATED USING STANDARD AIR-ALCOHOL VAPOR MIXTURES, AND BLOOD-ALCOHOL CONCENTRATIONS WERE COMPARED TO BREATH-ALCOHOL VALUES USING A BLOOD-BREATH RATIO OF 2,100 TO 1. THE DIFFERENCES BETWEEN THE BLOOD-ALCOHOL ESTIMATES FROM BREATH ANALYSIS AND THE ACTUAL BLOOD-ALCOHOL DETERMINATIONS VARIED, HOWEVER, DEPENDING ON THE ALCOHOL METABOLISM PHASE; BLOOD WAS LOWER THAN BREATH DURING THE ABSORPTION PHASE AND HIGHER THAN BREATH DURING THE ELIMINATION PHASE. THE ACCEPTED BLOOD-ALCOHOL TO BREATH-ALCOHOL RATIO (2,100 TO 1) IS TOO LARGE DURING ALCOHOL ABSORPTION AND TOO SMALL DURING ALCOHOL ELIMINATION. THE 95 PERCENT CONFIDENCE LIMITS FOR PREDICTING BLOOD ALCOHOL FROM BREATH ANALYSIS AFTER THE PEAK CONCENTRATION HAD BEEN REACHED WERE 0.084 MILLIGRAMS PER MILLILITER AT BREATH-ALCOHOL CONCENTRATIONS OF 0.50 OR 0.80--THE LEGAL LIMITS ON ALCOHOL CONTENT FOR DRIVING IN SWEDEN AND MANY OTHER COUNTRIES. TABULAR DATA, CHARTS, AND REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED--DAG)