This study evaluated the reliability of Widmark calculations, based on breath ethanol reading (BrACs), for estimating the amount of alcohol consumed. A standard ethanol dose (males 0.51 g/kg; females 0.43 g/kg) was given to 115 college seniors, and BrACs were measured for two hours. Calculations of ethanol dose were performed using BrACs taken at 60, 75, 105, and 125 minutes after drinking. Mean calculated ethanol doses were lower than actual doses at each time point (P < 0.001). Mean underestimates were 13, 12, 15, and 14 mL of 100 proof vodka at 60, 70, 105, and 125 min after drinking. Calculated doses overestimated actual doses in 11, 10, 3, and 3 subjects at 60, 75, 105, and 125 min after drinking. The maximum overestimates were 13, 11, 6, and 8 mL of vodka at 60, 75, 105, and 125 min after drinking. At the 95% confidence level, the calculated dose at 105 and 125 min did not overestimate the true dose, but could underestimate it by as much as 30 mL vodka.