Also evaluated were the accuracy and precision of the analyzers in self-monitoring of breath alcohol concentrations by comparing the results obtained by six volunteer drinkers with Breathalyzer Model 900A and Intoxilyzer Model 4011AS results. Studies with the simulator showed that Alcohol-Checker and Breathalyser PMT-1 readings were very sensitive to variations in the flow rate and duration of air flow (and hence, to variations in volume of breath exhaled). The Alcohol-Checker was also affected by the distance of the sensor from the outlet of the simulator. The Alco-Sensor III was unaffected by these variables. It was the only analyzer found to be sufficiently accurate, precise, and reliable over the clinically significant range of alcohol concentrations. Linear regression analysis of the results obtained in tests of drinking volunteers showed that Alco-Sensor readings correlated well with those of the Breathalyzer (r=0.982), whereas those obtained with the Alcohol-Checker and the Breathalyser PMT-1 correlated poorly (r=0.086 and -0.127, respectively). (Publisher abstract modified)