Two commercially available computer models were compared for reliability and validity in predicting the peak blood alcohol concentration (BAC) in 113 males and females, ages 21 to 81. All subjects consumed a constant amount of alcohol (30 grams) prior to undergoing breath testing to determine each person’s peak BAC. The observed BACs were then compared to the BACs estimated with each computer model. Both computer models seriously underestimated the observed BACs for these subjects. The results are discussed with particular reference to the need for additional studies of age, gender, and body composition as predictors of peak BACs in heterogeneous subject groups.