THE BREATH-ALCOHOL DEVICE IS BASED ON GAS LIQUID CHROMATOGRAPHY WITH A FLAME IONIZATION DETECTOR. A SUBJECT BREATHES INTO THE INSTRUMENT AND A MIXTURE OF 40 PERCENT HYDROGEN AND 60 PERCENT NITROGEN CONTAINED IN A CYLINDER CARRY THE BREATH OR VAPOR AND ACT AS COMBUSTIBLE GAS FOR THE FLAME DETECTOR. TWO ALTERNATIVE MODES ARE POSSIBLE: AN INTEGRATION MODE, IN WHICH ONLY THE ETHANOL PEAK IS ELECTRONICALLY INTEGRATED, THE INSTRUMENT RESPONSE BEING IN PEAK AREA UNITS; AND THE DIFFERENTIAL MODE, IN WHICH A CHROMATOGRAM OF ALL SUBSTANCES IN THE SAMPLE IS OBTAINED. CALIBRATION OF THE INSTRUMENT IS ACCOMPLISHED BY USING STANDARD ETHANOL VAPOR MIXTURES, AND THE ETHANOL CONCENTRATION IS CALCULATED USING PUBLISHED DATA FOR THE PARTITION RATIO OF ETHANOL BETWEEN AIR AND WATER. IN THE LABORATORY EXPERIMENT, ANALYSIS OF REPEATED SAMPLES OF KNOWN CONCENTRATION WAS DETERMINED BY MAKING 30 CONSECUTIVE SAMPLES OF THE HEAD-SPACE VAPOR ABOVE A STANDARD ETHANOL SOLUTION. THE ALCOHOL VAPOR CONCENTRATION WAS CHOSEN TO BE EQUIVALENT TO BREATH FROM A SUBJECT WITH BLOOD-ALCOHOL CONCENTRATION OF 0.083 MILLIGRAMS PER MILLILITER. THE RELATIONSHIP BETWEEN INSTRUMENT RESPONSE AND THE ALCOHOL CONCENTRATION IN THE VAPOR WAS CALCULATED BY MAKING MULTIPLE ANALYSES OF STANDARD ETHANOL VAPOR MIXTURES COVERING A RANGE OF CONCENTRATIONS, 10 DETERMINATIONS EACH, COMPRISING A TOTAL OF 70 ANALYSES OF ETHANOL VAPOR USING BOTH THE HEAD-SPACE AND SIMULATOR METHODS. IT WAS FOUND THAT A BREATH-ALCOHOL DETERMINATION MAY BE MADE WITHIN 2 MINUTES USING THE GAS CHROMATOGRAPH INTOXIMETER AND THE ETHANOL RESPONSE IS RESOLVED FROM POSSIBLE INTERFERENCE OF METHANOL, ACETONE, AND ACETALDEHYDE. THE DETECTOR SIGNAL IS LINEARALLY RELATED TO ETHANOL CONCENTRATION OVER THE PRESCRIBED BLOOD-ALCOHOL LEGAL LIMITS FOR DRIVERS IN SWEDEN. TABULAR DATA, CHARTS, AND REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED--DAG)